F._
                                                                  


.v

Paul A. Hanle, Editor
High Technology
on Earth: Studies
in Using Aerospace
Systems
and Methods

                                          '    *•	V



SMITHSONIAN STUDIES IN AIR AND SPACE - NUMBER 3

SERIES PUBLICATIONS OF THE SMITHSONIAN  INSTITUTION
     Emphasis upon publication as a means of "diffusing knowledge" was expressed by the
first Secretary of the Smithsonian. In his formal plan for the Institution, Joseph Henry
outlined a program that included the following statement: "It is proposed to publish a
series of reports, giving an account of the new discoveries in science, and of the changes
made from year to year in all branches of knowledge." This theme of basic research has
been adhered to through the years by thousands of titles issued in series publications
under the Smithsonian imprint, commencing with Smithsonian Contributions to Knowledge
in 1848 and continuing with the following active series:
Smithsonian Contributions to Anthropology
Smithsonian Contributions to Astrophysics
Smithsonian Contributions to Botany
Smithsonian Contributions to the Earth Sciences
Smithsonian Contributions to the Marine Sciences
Smithsonian Contributions to Paleobiology
Smithsonian Contributions to Zoology
Smithsonian Studies in Air and Space
Smithsonian Studies in History and Technology
      In these series, the Institution publishes small papers and full-scale monographs that
report the research and collections of its various museums and bureaux or of professional
colleagues in the world of science and scholarship. The publications are distributed by
mailing lists to libraries, universities, and similar institutions throughout the world.
     Papers or monographs submitted for series publication are received by the Smithsonian
Institution Press, subject to its own review for format and style, only through departments
of the various Smithsonian museums or bureaux, where the manuscripts are given sub-
stantive review. Press requirements for manuscript and art preparation are outlined on the
inside back cover.
S. Dillon Ripley
Secretary
Smithsonian institution

SMITHSONIAN     STUDIES      IN     AIR     AND     SPACE     •      NUMBER     3
High Technology on Earth
Studies in Using Aerospace
Systems and Methods
Paul A. Hanle
EDITOR

SMITHSONIAN INSTITUTION PRESS
City of Washington
1979

ABSTRACT
Hanle, Paul A., editor. High Technology on Earth: Studies in Using Aerospace Sys-
tems and Methods. Smithsonian Studies in Air and Space, number 3, 58 pages, 1 fig-
ure, 1979. —This collection of case studies describes how engineers or managers
applied their arts to five complex problems on earth. A variety of aerospace methods
and hardware were employed. Two of the cases concern "hard" technological develop-
ments: the construction of liquefied natural gas tankers and the creation of an auto-
mated system to transport people in Morgantown, West Virginia, both investigated
by Susan Frutkin. Three cases address aerospace software and methods of analysis:
the use of a technique of prediction developed by defense consultants, called "Delphi,"
by which opinions are collected and a sort of consensus is induced, investigated by J.
Gordon Milliken; the use of mathematical modeling in a computer to simulate the
flow of financial securities; and an attempt to reform some of California's public
services through aerospace systems analysis, both investigated by Carole R. Cristiano.
The focal point of each study lies in analyzing the political, technical, and bureau-
cratic forces at work in developing a complex system. Each study is summarized in a
separate, short comment, which also discusses how conflicts of goals and judgment
conditioned the outcome of the development. The editor's critical introduction to the
entire work places the five studies in a context of contemporary writing on technology
and society.
OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded in the
Institution's annual report, Smithsonian Year.
Library of Congress Cataloging in Publication Data
High technology on earth.
(Smithsonian studies in air and space ; no. 3)
Bibliography: p.
1. Technology —Addresses, essays, lectures.    2. Technology —Social aspects —Addresses, essays, lectures.
  I.   Hanle, Paul A.    II. Series.
T185.H53    600    78-14329
  
Contents
INTRODUCTION, by Paul A. Hanle		1
POLITICS AND THE PEOPLE MOVER, by Susan Fmtkin		9
LIQUEFIED NATURAL GAS TANKERS, by Susan Frutkin		19
THE DELPHI METHOD OF TECHNOLOGICAL FORECASTING,  by J. Gordon
Milliken		30
COMPUTER SIMULATION AND THE SECURITIES INDUSTRY,  by Carole  R.
Cristiano		44
SYSTEMS ANALYSIS IN THE CALIFORNIA STUDIES OF PUBLIC SERVICES, by
Carole R. Cristiano		53



High Technology on Earth
Introduction
Paul A. Hanle

  A few words are in order to explain what is at-
tempted in this book. We should begin with a dis-
claimer: it is not a litany of benefits from the aero-
space industry, nor is it a study of the impact of
"technology transfer" as the National Aeronautics and
Space Administration (NASA) and other federal
agencies have fostered it. Rather, this small collection
of case studies describes how engineers applied aero-
space arts to five complex problems on earth, each of
which entailed dealing with an array of hazards and in
which were employed a variety of aerospace ap-
proaches. The cases range from the development of
liquefied natural gas tankers to the reform of some of
California's public services. In addition, they include
the construction of a "people mover" in Morgantown,
West Virginia; the use of mathematical modeling to
simulate the flow of securities; and the use of a tech-
nique called "Delphi" for collecting expert opinions
and inducing a sort of consensus. To be truthful, none
of these studies document the history of problems that
have been completely "solved," although there were
several impressive technical successes.
  The focal point of each study lay in analyzing the
forces at play in developing a complex system, not
merely in pronouncing each attempt a success or a
failure. To accomplish this analysis would mean to
learn a little of how to use other high technological in-
ventions, and this was the purpose of our research. We
have concluded that it is always difficult but still pro-
ductive to attempt judiciously to use aerospace inven-
tions for other purposes. Not only what is judicious,
but how that is determined is the crux of each of the
cases presented below.
Paul A. Hanle, Department of Science and Technology, National
Air and Space Museum, Smithsonian Institution, Washington, D.C.
20560.
  
The following studies describe managerial methods
as well as technological "hardware," the approach of
systems analysis, and the discipline and principles of
aerospace engineering. Encompassing the processes of
using these, a suitably general but definite term was
sought to motivate the title of this book, for each of the
usual terms was in some way unsuitable. To scholars of
international economic and technological develop-
ment, the familiar but ambiguous "technology trans-
fer" means international aid: sending bulldozers to
Iran and teaching villagers in Upper Volta to drill
deep water wells. "Fringe benefits of space" and
"spin-off are associated with NASA, but much of the
technology described below comes as well from work
done on federal defense contracts, a circumstance that
occurs again in a larger sample of cases than presented
here.1 (The reason is that the source of much inven-
tion resides in the aerospace contractor, who has both
NASA and defense work, not in the contracting agen-
cy, which may wish to take the credit.) "Technology
utilization," though homely, comes closest to describ-
ing the situation. Unfortunately, in normal aerospace
usage it also embraces the profitable use in flying
machines of aerospace technology— the technical ben-
efits of the "Landsat" Earth Resources Satellite, for
example— and these are excluded from this book.
  So we have settled on a less elegant phrase—using
on earth the technology of the air and space industry,
or simply aerospace technology on earth. There is no
difficulty in denoting the diverse elements of aerospace
expertise with the term "technology," because tech-
nology has in recent years come to mean the applica-
tion in general of knowledge for practical ends, and
that application may be through discipline, principle,
methods, hardware, or people.
Such concern for a proper title may seem gratui-

2

SMITHSONIAN STUDIES IN AIR AND SPACE



tous. Still, if one does not begin on the right foot, one
soon stumbles in this obstacle-strewn field. Like the
distinction between industrial invention (creating an
original thought or device) and industrial innovation
(changing that which is produced for sale or use,2 the
distinction between building air and space vehicles
and applying the expertise gained thereby is crucial,
yet it is often muddled in lay and professional minds
alike. Furthermore, it is both transfer and diffusion of
technique that we consider here, a broader sweep than
is properly covered by the terms we have discounted.
Research for the National Air and Space Museum
  In early 1972, the administration of the National
Air and Space Museum began to plan a permanent
exhibition on the "Benefits from Flight." As one of the
major exhibitions of the Smithsonian Institution,
"Benefits" received congressional funding of one-half
million dollars and required four years to complete.
The exhibition opened with the new museum on 1 July
1976. Two and one-half of those years had been
devoted to research, documenting instances of the use
of aerospace technology on earth. From the hundreds
of case studies that were pursued initially by re-
searchers working in the museum, a number of which
were researched subsequently by hired experts, the
editor selected the five studies this book comprises.
After the cases were chosen, authors Frutkin, Milliken,
and Cristiano researched them additionally to take ac-
count of events through September 1975; they then
rewrote them to conform to the narrative format.
  The question remains as to why such an intense
program of research was undertaken in the first place.
We can answer only by reviewing how the "Benefits"
gallery was conceived. A proposal of 1973 to Congress
for funding this exhibition and research suggested
ambitiously that "... 'Benefits from Flight' can serve
as a powerful and productive channel for this sorely
needed technology transfer process." 3 As the curator-
ial staff (which included two principal investigators
before the editor) delved into the literature, they
found increasing need to document thoroughly the
cases of benefit to mankind that were to be presented
in the gallery. It was not enough to list the "dividends"
of aeronautical and space research in the sort of roll
that administrators of federal agencies, probably to
their distaste, found necessary each year to recite.4 For
the purpose of a major exhibition, which would bear
the Smithsonian imprimatur, be visited by five million
or more persons each year, and ideally foster more in-
novation, narrative stories had to be founded on re-

search of a depth and style theretofore not under-
taken.
  If this was not obvious at first, it became so after the
hired investigations got underway. Many instances of
applying aerospace technology had not been re-
searched fully,5 and each of the cases investigated by
the researchers seemed to be governed by its own set of
rules. Soon they concluded that the scope of the term
"technology transfer" was too limited—it was not a
process, but many processes, each of which seemed to
be defined only by the case that it described. These
processes were entwined on each other so that they
became harder to follow the further one progressed
from the starting point. Indeed, there seemed to be a
principle of frustration at work, in which if one began
with a "benefit" (a device, system, or even an idea of
immediate use), its connections to aerospace industry
were lost in a mist of subsequent commercial develop-
ment; or if one began with the aerospace invention, its
application on earth appeared to sink into a pool of
further invention and modification. It was nearly im-
possible in either case to trace the thread of "transfer,"
because, as the researchers finally accepted, no single
thread existed.
  One well-known social commentator recognized this
fact several decades ago. In his innovative study of
1946, The Social Effects of Aviation, William F.
Ogburn spoke of the reasons for the difficulty of trac-
ing derivative effects of inventions, reasons that our
investigators found to be valid in many of the cases
they studied.
We have noted that these derivative influences do not operate alone,
but join with other influences, and that the derivative influence of
any one invention seems to thin out and lose force. Furthermore,
these derivative influences fan out in various directions. It therefore
becomes rather difficult to isolate, trace out, and measure the deriv-
ative influences of an invention.
"But," concluded Ogburn, "these influences are real
and produce social changes, despite the fact that they
are difficult to measure." 6 While the existence of
derivative effects was not in question, either by
Ogburn or by the museum's researchers, the latter
came to Ogburn's conclusion and had to settle for a
description of the facts of the cases as they could find
them, eschewing as much bias as possible. Even this
little information, the researchers and the editor
hoped, would hold some lessons for how people inno-
vate with aerospace technology on earth. The cases
exhibited in the gallery and the five cases presented
below were studied partially to this end.

NUMBER 3

The Reasons for Selecting These Five Studies
  From a field of 21 unpublished, contracted re-
searches, the editor wished to choose a sample of
studies of reasonable length, which would represent
the highest quality and the most interesting relation-
ships of humans to technology while reflecting the
diversity of the entire research program. As it hap-
pened, the criterion of quality was applied with
greater rigor, and thus the distribution of studies has
been skewed toward certain topics: transportation sys-
tems, management, systems analysis, and prediction.
Although medicine, public service, energy, computer
analysis, and other topics also were investigated (and
were treated in the exhibition), the majority of this re-
search is not presented here. Quality and broad inter-
est have been represented at the cost of heterogeniety.
  Though obliged to state this, we feel no need to
apologize. In the following studies, circumstances and
events appear that are as representative of all attempts
to use aerospace technology as in those that have not
been published. Indeed, if the purpose of this collec-
tion were only to aid specialists in each industry repre-
sented, there would be little reason to bind them
together in one volume.
  But we can make more stringent claims on this col-
lection and so answer the question of why one might
want to read them together— that is, why we have cho-
sen to compile this anthology at all. In short, the cases
were chosen for their illustration of general principles,
not for their speciality. For example, while Susan
Frutkin's story of the Morgantown "people mover"
reveals some of the workings of the Boeing Company,
it more importantly portrays the need to perpetually
reconcile social and technological interests in massive
government programs. The development of liquid
natural gas tankers points to a common, perhaps the
most important, method of industries reusing aero-
space technology—the physical transfer of persons.
Gordon Milliken's study of Delphi methods suggests
the uses and limitations of some techniques of indus-
trial planning that have been borrowed from air and
space industries. Carole Cristiano's studies of simulat-
ing the flow of financial securities on a computer and
of simulating four urban services of California hold
similar lessons: the problems of systems of society may
be too complex to model yet or, worse, they may be
impossible in principle to reduce to computer analy-
sis.7 The editor has joined the authors in hoping that
these five instances may be understood in their de-
tails, but beyond this that they may provide a body of
evidence to support a general view of more universal

interest,  a full-systemic approach,  to which we now
turn.
A Hidden Presumption in Much of the Literature
of "Technology Utilization"
  In an unpublished draft of comments on using aero-
space technology, Louis L. Bucciarelli, the National
Air and Space Museum's former curator of science and
technology, has questioned a recurring presumption
in the literature of technology that we too shall wish to
examine. It is that the development of new technolo-
gy, like the machine, runs by itself, independently of
the human spirit. He observes:
  Those concerned with the maintenance of a viable economy assay
the accomplishment of technological innovation and dispersion.
"Technology Transfer" and "Utilization" are their key words. These
persons see "barriers" to the dispersion of technique; talk in terms of
time lags—duration from invention to widespread use—and recom-
mend policies to shorten this period. They trace out in hindsight a
sequence of related events, improvements in technique, and scientif-
ic discovery that have converged on a new hybrid corn, a heart
pacemaker, a solid state component.
  Behind this endeavor is the unstated presumption of technology
as "thing in itself." Technology as "black box"—a self-contained
unit that one builds, plugs in, and reaps the benefits of. Technology
as something out there to be discovered and developed to meet hu-
man needs upon demand. Technology as having an independent
existence like the law of universal gravitation, as some perceive it.
  To suggest that technology be considered as a cultural artifact,
like art, literature or music is to view the matter from a quite differ-
ent, more fruitful, perspective. Who would attempt to transfer our
most popular literature to a less developed country for its self-
improvement?
Following Bucciarelli, we invite the reader to examine
the connection of technology to the rest of human en-
deavor in the five case studies that follow. We ac-
knowledge from the start that they suffer also from a
deficiency of humanity, but what they indicate is more
than has appeared before.
  Bucciarelli criticizes a view that has been expressed
often in the years since the mid-1960s. As electronic
automation and control played an ever more promi-
nent role in society, Lewis Mumford and Jacques Ellul
projected the image of technology out of human con-
trol.8 They legitimized it to a great extent, even if, as
Langdon Winner has shown, it had its historical and
literary precedents.9 Universities created programs of
"Technology and Society" in part to investigate their
charges. In sociology, politics, history, and philoso-
phy, scholars began to explore the manifold relations
of humans to their technological and material culture.
Now the products of their research are beginning to
reinforce the view that we must understand technolog-
  
SMITHSONIAN STUDIES IN AIR AND SPACE

ical development as part of a larger structure of hu-
man social and political development. These scholars
have, in their holism, surpassed the critiques of Mum-
ford and Ellul. We shall return shortly to who they
are and what they have to say.
  Why must man and technology be understood in
their interaction, rather than as independently func-
tioning components, and how does this apprehension
pose problems that programs of technology utilization
do not address? It is because valuation of technology is
a key ingredient, because it is a measure of technologi-
cal success, the link between mankind and technology,
and yet the element of analysis that advocates of "util-
ization" pay little attention to. If we insist that the goal
of federal technical programs, or example, is to create
the best quality of living for the most people, then we
may indeed find that these programs move us toward
the goal with purely technological or economic suc-
cesses, ends that those who seek "utilization" pursue.
The values are firmly established, if unstated. Some-
times we find, however, that the welfare of society
.is best served by renouncing these narrower forms of
"success" when they threaten goals or needs that we
rank higher on a scale of value. Thus nuclear breeder
reactors have not been developed as fast as was once
expected, partly because of growing fear that they are
unsafe, first among environmentalists but increasingly
among others who stand to be endangered (and bene-
fitted) by them.10 Many people of reasonable mein
now agree that the danger outweighs our need for
more electric power— from this source at least. So also,
the aeronautics and space energy sponsored an expen-
sive, impractical "sight switch" to direct a motorized
wheelchair, and after the device had proved its techni-
cal feasibility in prototype, its manufacturers ceased
production, because they could not make it work un-
der "real" conditions.11 NASA promoted the device, it
is fair to say, beyond the point where reasonable ex-
pectations of practical return had withered, and it
died in a jungle of competing values. Neither NASA
nor the energy agencies considered these values, for
they were not thought to be part of the "technical"
problems the agencies were created to handle.
  Are these examples of failure of technology or of
failure of people? From these failures, wherever we
place them, can we draw lessons for the future use of
aerospace technology? Author Robert Pirsig has sug-
gested one answer to these questions, which we find
most interesting and in close alliance with the views of
Ida Hoos,12 whom we quoted above, and several
others. Pirsig has put forth his view in the form of a

perceptive story titled Zen and the Art of Motorcycle
Maintenance. He has much to say, we think, to pro-
moters of aerospace technology. Pirsig explains that
the way to understanding how we value technology lies
in our analyzing the relationship of humans to tech-
nology and not in bestowing absolute traits on either
one.13 Regarding what many see as ugliness in tech-
nology, he asserts:
But the real ugliness of modern technology isn't found in any mate-
rial or shape or act or product. These are just the objects in which
the low Quality appears to reside. It's our habit of assigning Quality
to subjects or objects that gives this impression.
  The real ugliness is not the result of any objects of technology.
Nor is it, if one follows Phaedrus' metaphysics, the result of any sub-
jects of technology, the people who produce it or the people who use
it. Quality or its absence doesn't reside in either the subject or the
object. The real ugliness lies in the relationship between the people
who produce the technology and the things they produce, which re-
sults in a similar relationship between the people who use the tech-
nology and the things they use.14
Much modern technology—the trashy plastic utensils
and the "lemons" of the assembly line —bears the onus
of "low Quality," not because that quality resides in
the things themselves, but because it is reflected in the
often mean, uninspired manner in which they are con-
ceived and used. We must evaluate technology in the
relationship between the object and its user. This rela-
tionship reflects that between object and maker, and
to "value" technology is to judge that relationship to
be good.
  To those who maintain that what Pirsig calls the
relationship of subject and object rests solely in the
technology itself or in the persons who use it, micro-
biologist Rene Dubos, allied with Pirsig on this point,
offers a genetic metaphor. Dubos writes of "important
problems of life," though his musing is evidently in-
tended to exceed the bounds of biological science:
  Present scientific fashions notwithstanding, there is no valid rea-
son for believing that the most important problems of life concern
its analysis in terms of genes, their subunits, or the chemical re-
actions they control. Far more important, it might be argued, are
the complex interrelationships between living things and their
total environments. Civilizations are generated by such interrela-
tionships. Life is so profoundly influenced by the evolutionary, ex-
periential, and social past that even a highly sophisticated physico-
chemical approach leaves out of consideration most of its deter-
minants and manifestations. "We murder to dissect," Wordsworth
said in "The Tables Turned."
  Since the living experience disappears when the organism is taken
apart, many aspects of human life can be understood only by study-
ing man's functioning with all its complexities and in the responses
that he makes to significant stimuli. Such a study would require an
organismic and ecologic attitude very different from the analytic
one which now prevails in biology.15
 
NUMBER 3

The wisdom is age-old. Dubos notes that the Hindu
mystics who composed the treatises called the "Upan-
ishads" pointed to such an organismic attitude 2500
years ago. It applies as well to mankind's use of tech-
nology.
  We have cited Mumford as one who projected the
image of the soulless machine of technology; now we
must again take care to distinguish between his atti-
tude and those of Pirsig and Dubos. Mumford would
place the source of human alienation from technology
in the "systematic dissociation [of work with ma-
chines] from the rest of life." 16 He views the develop-
ment of power-harnessing machines as a curse, and so
his attitude epitomizes that of a group often identified
as "anti-technologists." He believes that mankind's
technological burden has accrued to it because it has
allowed technology to dominate. Mankind (the race)
and technology (the object) are alienated from each
other, but this is a consequence of something intrinsic
in the technology itself. Pirsig and Dubos, however,
condemn less the machines than those who would
blame some intrinsic evil for their domination of our
lives. Dubos raises a call to "make technology subser-
vient to worthwhile human needs," 17 and this sounds,
as one recent critic has observed, like a reaction
against Mumford's self-determined machines.18 In
Dubos's view, however, holism bears emphasis, for only
in the whole of man and environment is carried the
basis by which we value technology, the element of
human relationship to technology in production and
use. So better we might ally Dubos and Pirsig, and
we think that their views are substantiated in the cases
that follow.
  Let us conjure one last writer, the recent critic
Samuel Florman. A successful engineer and a reason-
ing technocrat, Florman has described some of the
rewards and frustrations of building and using ma-
chines. His analysis of the views of the antitechnolo-
gists proceeds from their writings, and he concludes
that their collective viewpoint resembles that of
Jacques Ellul. "A primary characteristic of the anti-
technologists is the way in which they refer to 'tech-
nology' as a thing, or at least a force, as if it had an
existence of its own." 19 His and Bucciarelli's criticisms
are similar, though he addresses the antitechnologists;
Bucciarelli comments on the promoters of technology
transfer. If the two are to be identified, as the quota-
tion suggests, we leave the unsavory task to the reader.
  Unfortunately, Florman has linked Dubos's poetic
imagery closely to Mumford's and Ellul's more pessi-
mistic expressions.20 But on balance Florman has sum-

marized adequately a major element of the movement,
and he has attempted to show where it has gone awry:
But sober thought reveals that technology is not an independent
force, much less a thing, but merely one of the types of activities in
which people engage. Furthermore, it is an activity in which people
engage because they choose to do so. The choice may sometimes be
foolish or unconsidered. The choice may be forced upon some
members of society by others. But this is very different from the con-
cept of technology _£.e//~misleading or enslaving the populace."
He summons philosopher Daniel Callahan to deny the
dualism of man and technology and to replace it with
a view of technology as but one of the manifestations
of man:
At the very outset we have to do away with a false and misleading
dualism, one which abstracts man on the one hand and technology
on the other, as if the two were quite separate kinds of realities. I
believe that there is no dualism inherent here. Man is by nature a
technological animal; to be human is to be technology. If I am cor-
rect in that judgment, then there is no room for a dualism at all. In-
stead, we should recognize that when we speak of technology, this is
another way of speaking about man himself in one of his manifesta-
tions."
  On this point, finally, it is the conclusion of writers
like Pirsig, Dubos, Florman, and Bucciarelli that ear-
lier critics of "technology" have erred in choosing their
objects of criticism. Even the masterful Lewis Mum-
ford and Jacques Ellul have tended to elevate "hard-
ware" above mankind's acts of creation and use of
these tools. Instead, we must look more closely at the
relationship of mankind and technology. When we ex-
amine this relationship, we look for the values "within
it. Our conclusion is not so important as the process of
studying the relationship —of finding the values —be-
cause in that process we identify social forces arising
from these values, which play against each other and
determine the direction of our technological society.
Such a study, which is hardly begun in the cases that
follow, may hold lessons for the future of aerospace
technology. We think that our approach, at least, is
how those writers whom we have cited might deal with
our problems. Our goal is to understand those social
forces. If the studies fall short of the mark, it is
because our effort has been insufficient, not because
the method is wrong.
Finding the Values
  How might one evaluate explicitly the relationships
of people to technology? We have already noted that
elements other than the economic are involved when
we look at each case. (We emphasize this because to a
large extent, it seems, economics motivate people to
  
SMITHSONIAN STUDIES IN AIR AND SPACE

develop new technology in general.) 23
  Technical elegance alone may bring the relation-
ship to flower. In the history of science there are few
clearer instances of a researcher fashioning a symbiotic
relationship with his experimental apparatus than that
of physicist A. A. Michelson. In almost five decades of
research around the turn of the century, he nursed the
instruments he invented, coaxing from them both pre-
cision and high reliability. In his duplication of
Foucault's measurement of the speed of light, then
later in his much-repeated interferometric determina-
tion of the constancy of the speed of light (propagat-
ing perpendicular or parallel to motion of the source
and detector) , the legendary precision of Michelson's
experiments was unequaled. When his scientific suc-
cessor Dayton Miller later attempted the same experi-
ment with a much larger interferometer at the top of
Mount Wilson, his conflicting results caused some con-
cern, but physicists finally discounted them. Accord-
ing to visiting theoretical physicist Max Born at least,
the source of Miller's error was to be found in the
unreliability of his elephantine machine.24 In the
terms used above, we can say that Miller failed to
evolve the quality of the relationship between him and
his experimental apparatus that characterized Michel-
son's research.
  Another element of technological value is shared
with science. It is that the maker and thence the user
may feel a sense of joy in creating and manipulating
the invention —in bringing it to function as it was in-
tended. The feeling is that of sport or play, and it has
been discussed by many scholars, most notable among
them the psychohistorian Lewis Feuer.25 While some
critics dispute his claim that the rise of modern sci-
ence in the 17th century was due to "the spirit of play,"
we note that he has identified an element of research
that also motivates engineers. "Technical and scien-
tific work is usually fun. In fact, creative technical
work provides much the same satisfaction that is ob-
tained from painting, writing, and composing or per-
forming music," affirms President Jerome Wiesner of
the Massachusetts Institute of Technology.26 The
engineer solves his technological puzzles as intently as
others decipher cryptograms; he designs his machine
as creatively as a composer writes a concerto. The
value of a masterpiece accrues to him who rejoices in
the act of creating it as much in engineering as in
musical composition. And we, the users, if we are in a
sympathetic frame of mind, value it also for the joy
and skill of its creation. For some, there is the mere in-
spection  of an   antique  airplane  or  automobile,   a

graphic print, or the structure of a bridge; for others,
it is the bond of human and machine, the life-and-
death dependence of a driver on his racer that brings
joy and a keenness to his senses.
  We may find that competition enters into the way
we ascribe value. Does it seem that advocates of the
United States supersonic transport placed much value
in surpassing the Russians and the French-English
consortium? The American project of the 1960s en-
tailed not just building such a transport but making it
the biggest and fastest of the lot, finally to the detri-
ment of the entire project.
  The element of control of purpose has come increas-
ingly into debate in massive "high" technological
developments, since the beginning of the recent era of
heightened awareness. Too often, it seemed, inter-
ested individuals or groups usurped our right and
obligation as users to measure the purpose to which
our technology would be put. One need but recall the
revelations of Ralph Nader, who in the mid-1960s ex-
amined the mechanical integrity of the Detroit auto-
mobile and declared it lacking. The industry, it ap-
peared, had allowed the technology of safety to take a
back seat to that of speed and mass production. To the
automobile business the purpose of the car was profit,
and it came at the cost of safety to passengers. As soon
as this fact was widely known and accepted (it is no
longer controversial, as major "recalls" are often cor-
porate admissions of past mistakes), the value of the
owner's relationship to his automobile diminished. As
if to rebuke Detroit, American consumers began to
purchase neatly manufactured, though rarely safer,
foreign models.27 The purpose of the builder-machine
relationship had been altered since the days of the
Model T, and with it the value that many had per-
ceived in domestic cars.
  Other elements of valuation may arise from periph-
eral sources. The secondary consequences of a person's
use of mechanical technique may alter his attitude
toward it. (Consider the case of Theodore Taylor,
whose preoccupation with the technical elegance of
the atomic bombs he created yielded to his growing
alarm that these devices could be home-built and used
for terror and extortion.) 28 The mere fact of success,
especially in a complex undertaking, can excite the
imagination. (Consider the Viking Mars Landers,
which brought no sign of extraterrestrial life to earth
by 1977, but which nonetheless caused a flurry of lec-
tures, popular publications, and general interest
among more than space enthusiasts.) And, of course,
there are others.
  
NUMBER 3

  Yet something lacks in such an analysis of the ele-
ments of valuing a technological relationship. Did we
not aver that one must look at the relationship holis-
tically and so not "murder to dissect"? In each case
cited, indeed in each that can be found, the elements
of value are determined by the greater culture of both
creators and users of technology. For Michelson, the
greater culture was unambiguously defined by those
who shared the values of physical science. For the
automobile industry, however, problems arose when
the elements of valuation in the world of business
diverged from those of the' nation of users. The in-
dustry's disregard for certain values of the users (like
passenger safety) engendered mischief.
  A lesson from this and other cases (like the com-
puter model of securities exchange discussed below,
which proved less cogent than expected because its
creators ignored conservatism and rivalries among the
exchanges) is that we must not underestimate social
aspects in a "technical" problem. Repeatedly, as we
shall see, it is those social aspects and their political
overtones —specifically the conflicts of value within the
entire culture that creates and uses the technology—
that precipitate the fall of a "technical" program. The
balance sheet that betrays the failure is only one meas-
ure of it and does not reveal the underlying social
forces that caused it.
  And yet we find in the cases to be presented that the
complexity of society often renders an "at once" ana-
lytical and unified view impossible. In four of the five
cases, problems occurred as much because the social
values were unidentifiable or undefined as because the
developers made no effort to grasp them. Who could
identify the cultural norms of 60 different agencies
with which the builders of liquid natural gas tankers
had to deal and each of which claimed to protect the
interests of society? Who was to say finally of the pur-
pose of a personal rapid-transit system in Morgan-
town, West Virginia, whether priority should be given
to solve traffic problems of that university city or to
demonstrate a prototype of similar systems to be built
throughout the country? Who was to say if the securi-
ties exchange industry was modeled to protect, re-
form, or replace the existing industry? Could one
possibly have considered all the conflicting values of

those who supplied and used the California services
that were analyzed? It seems sometimes that if such
endeavors work at all, it is because those who are in-
volved in them share an unstated will to succeed that
overpowers their often strongly stated conflicts of other
values.
  The examples suggest, then, as Ida Hoos would ar-
gue, that we cannot solve fundamental social "prob-
lems' through a purely engineering approach. Holism
is antithetical to analysis, and in any event the prob-
lems lie less in technique than in the conflicts of value
of "builders' and "users." It is the irrational in us that
sustains our values, and these values account for, in-
deed define, technological failure or success in society.
Perhaps those who develop technology need more cul-
tural vision, not more analysis, if they are to apply
their (and our) machines successfully to social ends.
  Our use of holism has its origins in organic meta-
phore, so we must underscore that we are likening the
relations of people and machines to those of orga-
nisms. This admission is not to fully preclude the pos-
sibility of rational analysis. C. P. Snow, on the con-
trary, noted that holism "has nearly always been a trap
for working scientists. DNA and the genetic code
would never have been unraveled by people who were
moved by the concept of holism."29 Instead, the
research scientist works on his narrow problem, di-
vorcing himself purposely from his surroundings so as
to gain a "controlled environment" in which he can
measure minor changes of the system he observes as he
varies each parameter of that system. Where this sepa-
ration perturbs only little the system being measured,
the scientist (or technologist) is on firm turf. But the
ground becomes soft when, as in many social "systems
analyses," the exclusion of social, political, and other
cultural influences cuts the heart out of his problem.
The point is that we are all social beings, more or less,
and a few are trained to reason as scientists or engi-
neers. It is reasonable to call for the rare abilities of
engineers who allow the living factors into their
studies. If by so doing, the methods of engineering are
weakened or rendered useless, then the analysis must
pass to other techniques. Not all are analytical, indeed
many are not even rational, but are proper ele-
ments of using aerospace technology on earth.
  
SMITHSONIAN STUDIES IN AIR AND SPACE
Notes

  1. See, for example, Hymen Olken, Technology Transfer: How
to Make It Work (Livermore: Olken Publications, 1972).
  2. See, for example, Erik A. Haeffner, "The Innovation Proc-
ess," Technology Review, (March/April 1973), pages 18-25.
  3. Major Exhibition Programs, Fiscal Year 1974, Smithsonian
Institution (budgetary document).
  4. See, for example, U.S. Congress, House, Committee on Sci-
ence and Astronautics, Hearings, on 1975 NASA Authorization
H.R. 12689, 93rd Congress, 2nd Session, (1974), part 1, pages 2,
16, 17, 22, 23, 71-73, 75, 76, 83-87, 182.
  5. One example: Public Technology, Incorporated, in prelimi-
nary research under a contract from NASA, tested a cold mix
street-patching material which "found to be the best formulation
after some of the early tests, appears to be demonstrating the kind of
permanence and durability cities and counties require," (ibid., part
4, page 28), according to President Porter W. Homer of Public
Technology, Incorporated. Yet, according to the subsequent study
performed for the museum by PTI Vice President Warren D.
Siemens, "the amount of improvement over the city winter mix of
the . EVA-modified cold mix was not [significant]" (unpub-
lished report).
  6. William Fielding Ogburn, The Social Effects of Aviation
(Cambridge: Riverside, 1946), pages 76-77.
  7. Ida Hoos has expressed this view with respectability and
vigor:
The engineer soon discovers that crime, welfare, and the like
are not one system, but many systems: not one problem, but
many problems, only some of which fall within any disciplinary
or "scientific'' scope. Indeed, for most social problems, there is
no technical or methodical "solution." Their very definition de-
pends on the point of view of the analyst; his value system is
imbedded in the way he designates the objectives, identifies
pertinent variables, and assigns costs and benefits. And in all
these crucial matters, his schooling and experience as an engi-
neer may hinder rather than help him. Thanks to the process of
"trained incapacity," he, probably more than any other profes-
sional, purposefully rivets attention on certain types of prob-
lems or aspects of them and does well what he does, with careful
disregard for those matters which he considers outside his pur-
view. (Ida R. Hoos, "Systems Analysis as Technology Trans-
fer, "Journal of Dynamic Systems, Measurement, and Control
(March 1974), pages 1-6, especially page 3.)
  8. See, especially, Lewis Mumford, The Myth of the Machine:
The Pentagon of Power (New York: Harcourt Brace Jovanovich,
1970) ; Jacques Ellul, The Technological Society (New York:
Knopf, 1964), translated by John Wilkinson. Ellul's book first ap-
peared in France in 1954 but attracted more attention after its
publication in the United States in 1964.
  9. Langdon Winner, Autonomous Technology (Cambridge:
MIT, 1977).

  10. Philip Handler, president of the National Academy of Sci-
ences, counsels caution: "Before such a program is implemented, it
must be examined in the most intimate detail and the public must
be satisfied that the risk to future generations is of an acceptable
order; that the latter is the case is not self-evident at this time."
Quoted from "On the State of Man," Interdisciplinary Science
Reviews, volume 1 (September 1976), page 196.
  11. According to NASA's Robert Zimmerman, formerly of Of-
fice of Technology Utilization, quoted in "Medicine's Inheritance

from Space," Medical World News (14 July 1975), page 110.
12. Hoos, "Systems Analysis."
  13. Robert M. Pirsig, Zen and the Art of Motorcycle Mainte-
nance (Toronto: Bantam, 1974), especially pages 131-132,
233-234, and 283-284.
14. Ibid., pages283-284.
  15. Rene Dubos, So Human an Animal (New York: Scribner,
1968), page 120.
  16. Lewis Mumford, "Technics and the Nature of Man," Tech-
nology and Culture, An Anthology (New York: Schocken, 1972),
Melvin Kranzberg and William H. Davenport, editors, page 209.
17. Dubos, So Human, page 231.
  18. Samuel C. Florman, The Existential Pleasures of Engineer-
ing (New York: St. Martin's, 1976), page 55.
19. Ibid., page48.
20. Florman continues:
  Making due allowance for poetic license, it is clear in these
repeated personifications that technology is considered to have
an existence separate and distinct from individual human
beings. Indeed, technology is thought of as something that,
unless fought against, can do things to human society, such
things as "claim priority," "take over" and "dictate," even
"ravish" and "murder." Dubos muses that although technology
cannot theoretically escape from human control, society feels
threatened by technological forces "just as it was threatened by
the raiding Norsemen and Saracens ten centuries ago."
  This way of thinking has spilled over into common usage, so
we are not surprised to see an advertisement that begins,
"Technology has trapped us . . . ," or an article in a news mag-
azine that says, "Technology is seen as a dangerous ally" (page
49).
21. Ibid., page 58.
  22. Callahan, "Proceedings of the Centennial Convocation of
the Thayer School of Engineering at Dartmouth College,'' held
23-25 September 1971, quoted in Florman, Existential Pleasures,
pages 58 59.
  23. See, for example, J. Schmookler, Invention and Economic
Growth (Cambridge: Harvard, 1966). Schmookler's hypotheses
have come under increasing criticism, but his ground-breaking
statistical study remains cogent.
  24. Born's evaluation is noted in D. M. Livingston, The Master
of Light (New York: Ch. Scribner's Sons, 1973), page 314. See M.
Born, The Born-Einstein Letters, translated by Irene Born (New
York: Walker, 1971), H. Born to A. Einstein, 11 April 1926, pages
89-90. The "aether-drift" experiments are described in Loyd S.
Swenson, Jr., The Ethereal Aether (Austin: University of Texas,
1972), pages 190-227, and in R. S. Shankland, "The Michelson
Morley Experiment," American Journal of Physics, volume 32
(1964), pages 16-35.
  25. Lewis S. Feuer, The Scientific Intellectual (New York:
Basic Books, 1963).
  26. Jerome B. Wiesner quoted by Florman, Existential Pleas-
ures, page 95.
  27. We make no claim that America's increased market for for-
eign cars was caused by Nadar et al., only that they contributed to
the growing American affection for the imported car.
  28. See John McPhee, The Curve of Binding Energy (New
York: Farrar, Strauss, and Giroux, 1974).
  29. Review of Louis J. Halle, Out of Chaos, in The New York
Times Book Review (24 April 1977), page 26.
  21. 
Politics and the People Mover
Susan Frutkin

EDITORS COMMENT
  This study of the 11-year history of a personal rapid transit sys-
tem, a "people mover," in Morgantown, West Virginia, is rife with
technical data that bespeaks the author's care in research, but it also
contains interesting analyses and judgments. Her story centers on
the interaction of three factions responsible for building the people
mover: the CongTess of the United States, which appropriated more
than $60 million for the test project; the administrators from the
Department of Transportation, who distributed the funds and man-
aged the entire program; and the engineering contractors and con-
sultants, who planned and built the system. Author Frutkin investi-
gates the politics of this interaction; and when one regards the
billions of dollars that the federal government distributes each year
in technological research, her political theme is of gTeat importance.
  Morgantown's problem in 1965, the frequent congestion of auto-
mobiles on the two main roads of the city, precipitated an alliance
between the administration of West Virginia University and a mem-
ber of the faculty, industrial engineer Samy Elias. Together they
sought funding for a system to transport students across the city be-
tween campuses several miles apart. Elias knew that there had been
no major changes in the technology of mass transportation in this
country in 60 years, nor were the older technologies suited to the
needs of Morgantown. He advocated developing a new system based
on high technology.
  For their part aerospace interests were involved for other reasons.
The Boeing Company, uncertain of future defense contracts sought
to diversify into mass-transit systems. At the same time, the Depart-
ment of Transportation, which funded the project, sanctioned hir-
ing the well-reputed Jet Propulsion Laboratory to manage design of
the system. Further stimulus to aerospace participation in the proj
ect came in 1970 from the Urban Mass Transportation Act, which
encouraged use of then economically depressed aerospace com-
panies. The evidence presented leaves no doubt that aerospace in-
dustry built the Morgantown project because both government and
the industry perceived that it could develop a new technical system
rapidly.
  The people mover at Morgantown was to be a demonstration of
new mass-transit technology. In early grants to West Virginia Uni-
versity, the Department of Transportation had insisted that the sys-
tem be developed with the intention of using the new techniques in
other cities; but allowing even for its status as a prototype, the proj-
ect had many problems. The preliminary phase of the program,
which included operating vehicles on an elevated guideway, was
Susan Frutkin, formerly with Booz Allen Applied Research,
Bethesda, Maryland, now with American Petroleum Institute,
Washington, D. C. 20037.

rushed to completion before the 1972 national elections. A break-
down of one of the five vehicles during dedication ceremonies in Oc-
tober was described by one bureaucrat with a melancholy justaposi-
tion of metaphors: "It was ridiculous . . . like putting five bread-
boards in the field." Later extensive tests showed that components
were defective, systems for safety and guidance were inadequate,
and computer electronics contained errors in logic. Installation of
the 45-car system was completed only in September 1975 with final
costs more than quadruple the original, no doubt unrealistic, esti-
mate. While many of the problems appeared to be technical in
nature, they often stemmed from underestimation of the time as
well as expense of developing a new system. Then too, the conflict-
ing wishes of the university, the Department of Transportation, and
Congress contributed to the feeling at the end of 1975 that none of
the three factions would be happy with the system. Nevertheless, the
Morgantown people mover is operating, and it carries people from
one campus of the University to another in that small city. Further-
more, the federal government planned in early 1977 to fund at least
three new people movers in larger cities. The Morgantown system
holds the status of prototype of these systems.
  Throughout the study, interviewees lamented that internal and
congressional politics were allowed to interfere with technological
development. Underlying their complaints is the manifestly vital
issue that those who develop technological systems wish to free them-
selves of "political" restraints, to solve their problems in isolation
from an untidy world that will not let them. Yet it is the untidy
world that poses the problems in the first place. So engineers can
only be lauded for bringing their problems into tractable dimen-
sions, yet their analysis is no less incomplete for doing so to the ex-
clusion of political concerns. A lesson of the people mover is that
developers of technological systems for society must learn to
"reckon" with political factors, indeed with many social factors that
cannot be quantified, for the systems derive their reason for being
from the benefit they provide a society where priorities are set in the
first instance by political means.
Introduction
  In the early 1960s, the town of Morgantown had a
population of about 22,000 residents and as many
West Virginia University students and staff. Morgan-
town runs along a river on steep hilly terrain; only two
streets completely traverse it. The University has three
main campuses: one downtown, the other two at op-
posite ends of town. While they are only 1.5 miles
away from each other, by the early 1960s, traffic jams
  
10

SMITHSONIAN STUDIES IN AIR AND SPACE



required 70-minute separations between classes on dif-
ferent campuses, and there were five periods of peak
travel.
  Dr. Samy Elias, who had earned a B.S. degree in
aerospace engineering and M.S. and Ph.D. degrees in
industrial engineering, joined WVU's industrial engi-
neering staff in 1965. For a number of years, he had
taught and conducted research on computer applica-
tions to urban transportation. After experiencing
Morgantown's traffic problems firsthand, Elias
became interested in the idea of a research project for
the engineering department to devise a new transpor-
tation system for the town. Over a two-year period, he
brought together the university's president, an engi-
neering staff, and an advisory committee of communi-
ty leaders and planners to study Morgantown's chronic
traffic problems.1
Proposal for a New Transit System
  Based on discussions and study of alternatives,
WVU and the advisory committee concluded that
Morgantown needed an innovative transit system.
They decided that the town's topography and popula-
tion distribution precluded widening highways or
building new ones. They also agreed that it was impor-
tant to minimize operating costs (the largest compo-
nent of which is labor) while providing maximum
service. Any new system would therefore have to be
either elevated or subterranean and would also have to
rely heavily on computerized automation.
  In 1967, WVU presented a proposal reflecting these
decisions to the Department of Housing and Urban
Development (HUD, the responsible agency at the
time) for federal funds to study the feasibility of an in-
novative urban mass-transit system. The proposal con-
tained several unique features. For the first time, it ad-
dressed the problem of a mass-transit system in a small
city. In addition, the system was to operate 24 hours a
day in response to different levels of demand. Further-
more, the proposal called for testing the system in an
urban setting instead of at the usual remote test track.
  Either because of the low priority placed on mass-
transit research and development at the time or from
sheer inertia, HUD did not act on the proposal for two
years. In the interim, however, impetus to new initia-
tives in this area came from the 1966 Reuss-Tydings
Amendment to the Urban Mass Transportation Act.
The major objective of this legislation was to apply ad-
vanced technology to mass-transit problems, and it
called for a program of research, development, and

demonstration of transportation systems for metropol-
itan areas of various sizes. The law eventually helped
generate interest within government and industry in
the development of automated guideway transit
(AGT) systems, among other new technologies.
  In 1968, the environment for conducting research
on and development of mass-transit systems was fur-
ther improved by the support of the incoming admin-
istration. The Urban Mass Transit Administration
(UMTA, created in 1964) was transferred to the re-
cently created Department of Transportation (DOT).
As Elias put it, "It was a new administration with a
new language."
  In May 1969, Elias visited Secretary of Transporta-
tion John Volpe to request consideration of his long-
pending proposal. Volpe liked the idea of testing small
vehicles operating on demand in a community, espe-
cially one like Morgantown, where the problems were
manifest and local initiative to solve them was evident.
In July 1969, Volpe approved $100,000 for the study;
WVU contributed $35,000.
  In about a year, Elias had completed the feasibility
study for a personal rapid transit (PRT) system, also
called a "people-mover." Using a 'systems" approach,
typical in aerospace work, he identified major subsys-
tems and analyzed their interrelationships. He also
studied construction and operating and maintenance
costs, along with such mechanical factors as high
speed, automation, and distances between cars.2 In
all, Elias considered about 130 PRT concepts before
selecting Alden, Varo, and Dashavayor (Bendix) to
do preliminary studies, for which DOT-UMTA
granted an additional $20,000.
  After WVU staff and an outside consultant evalu-
ated the preliminary engineering studies, they chose
Alden, a small nonaerospace company with two com-
puter-run prototype cars, to build a system featuring
6-second headways between cars.
  During this period, the WVU staff considered two
management approaches for design and construction
of the system: The university could hire 30 to 40
engineers in addition to its regular staff, or it could
hire a systems manager. Based on suggestions from
various people involved during the initial studies,
WVU contacted the aerospace-oriented Jet Propulsion
Laboratory (JPL) and Bendix Aerospace Company
about assuming the function of project manager. JPL
was selected on the basis of cost and its nonprofit
character, which was considered more appropriate for
a program with the potential national application of
PRT systems.3
  
NUMBER 3

11



  In August 1970, WVU submitted its proposal to
UMTA for a federal grant of $13.5 million to begin
designing the PRT system. Alden's initial request had
been for $25 million, but when UMTA said it could
not commit the full amount to one project, WVU
lowered its request to $13.5 million. Few who were in-
volved actually took the $13.5 million figure seriously.
Lack of staff, lack of understanding of what would be
involved, and the desire to present the PRT as a cheap
program were some of the factors that drove down ini-
tial estimates. Unfortunately, this approach was to
backfire and its consequences would plague the entire
project throughout its development.
  By mid-1970, however, congressional pressure was
increasing to apply advanced technology to national
problems. The Urban Mass Transportation Assistance
Act was passed, permitting DOT to provide capital
grants to any community for rejuvenating old or creat-
ing new mass-transit systems, including PRT systems.
The new law specifically encouraged use of depressed
aerospace and defense industries to develop new sys-
tems.4 West Virginia's Representative Harley Staggers
was also a strong and influential supporter of the Mor-
gantown program.
  UTMA, too, was now actively interested in conduct-
ing programs to demonstrate technical feasibility,
public acceptance, and cost effectiveness of an ad-
vanced PRT. The agency considered Morgantown a
good initial test site for acquiring demonstration data,
on the basis of which DOT could begin certifying sys-
tems for capital grants.5
Approval for the PRT
  In September 1979, UMTA granted $90,000 to
WVU to recommend design and to arrange for rights-
of-way of a 6-station, 120-vehicle PRT. Another $1.3
million was awarded to JPL to manage the system.
  Almost simultaneously, UMTA announced that it
would assume overall supervision of the project to en-
sure reproducibility and national applicability of the
system. Its first decision in that capacity was to ter-
minate Alden's contract on the grounds that Alden
had neither the financial resources nor the capacity to
carry out a large development project. (In fact, they
had only five full-time employees.) The agency also
pointed out that Alden would not qualify for the com-
petitive bidding because they had written the specifi-
cations." New specifications were drawn up and ap-
proved jointly by UMTA, WVU, and JPL.
On the basis of informal contacts made with indus-

try during this period, JPL prepared a rough project
cost estimate of more than $37 million, compared to
the initial $13.5 million. UMTA had, in fact, budg-
eted about $20 to $24 million for the demonstration.
Lacking detailed technical designs and faced with ris-
ing costs, UMTA told JPL to propose an alternative
system costing $20 to $30 million. A PRT system that
included five automated vehicles, guideway, three sta-
tions, software, tests and evaluation, and one mainte-
nance facility was determined "sufficient to address
UTMA's major program objectives of establishing
technical feasibility while limiting the program
cost." 7
  In April 1971, DOT solicited the transportation in-
dustry for proposals to design and build the system. A
number of aerospace companies responded. In May,
UTMA let contracts for various components of the
system to companies selected by JPL. The Being Aero-
space Company was to build the PRT vehicles. A ma-
jor aerospace company in its first venture into ground
transportation, Boeing teamed with Alden (whose
fundamental vehicle-steering design was retained).
Sundberg-Ferrar was selected to design the passenger
compartments.
  Bendix Corporation was chosen for the command
and control systems, with Systems Development Cor-
poration named subcontractor to Bendix for the soft-
ware. Frederic R. Harris, Incorporated, received the
contract for design of power and guideway systems as
well as for the stations.
  Although UMTA expected JPL to perform the role
of system manager and technical integrator, the sepa-
ration of management and design of the PRT's com-
plex elements reflected the traditional approach of
transportation systems, i.e., buying them piecemeal
instead of as a package. This proved detrimental to
the complex integration of the PRT's elements, for
which an aerospace type "systems" approach was
almost mandatory.8
  For its part, Boeing's PRT bid was an outgrowth of
the desire to diversify the aerospace company. They
were also well aware that their aerospace expertise
qualified them specifically for entering the field. They
joked that the Boeing-designed Lunar Rover Vehicle
(LRV) of Project Apollo was really the first PRT.9 In
fact, in every phase of their participation in the proj-
ect, Boeing's basic aerospace character was an impor-
tant factor.
  A test facility for the Morgantown vehicle was in-
stalled at the Boeing Aerospace Company's Kent
Space  Center  in  Seattle.   Throughout  the  project,
  
12

SMITHSONIAN STUDIES IN AIR AND SPACE



assignments of personnel reflected aerospace ex-
perience ; a Minuteman ground systems supervisor and
Houston Space Center manager (responsible for
Apollo engineering analyses) was ultimately selected
PRT program manager; the LRV project manager
and several LRV engineers were eventually assigned to
the Morgantown project; from Boing's military air-
craft program came the PRT products support engi-
neering manager; in addition, aeronautical engineers
were later attached to the Bendix group to manage the
subcontractor. At its height, Boeing's PRT team
numbered about 200 people—all of whom had some
aerospace background.
  Shortly after design contracts were awarded, fric-
tion developed between UMTA and JPL and between
UMTA and WVU. On the one hand, UMTA officials
were skeptical of JPL's cost and scope forecasts. JPL
and Elias had envisioned vehicles for eight persons and
7.5-second headways, but DOT wanted larger cars (to
justify the program to Congress) and 15-second head-
ways. DOT also decided against providing each car
with the capability of pushing a disabled vehicle off
the guideway. This would have required more sophis-
ticated steering and collision-avoidance designs,
costing server al million more.
  According to Elias, however, it seemed that "DOT
didn't have faith in the technology; they weren't tak-
ing advantage of the technology that was available. . .
JPL said there was nothing they couldn't do: they had
gone to the moon." 10
  For its part, DOT thought JPL was falling behind in
defining the stystem. "There was not enough of the
'Let's get moving' industry attitude," said one of the
administrators of the program.11 Under strong pres-
sure from Secretary Volpe to demonstrate the new
PRT before the November 1972 presidential elections,
UMTA was as anxious to start construction as they
were to minimize costs.
  After a year had elapsed without a contract, JPL
and DOT still could not come to terms, and UMTA
asked Boeing and Bendix to present proposals for
systems management of the project.12
  By mid-1971, disagreement between UMTA and
WVU had also increased, and the university was
eliminated from all management decisions, although
it continued in an "advisory capacity." UMTA drew
up a contract with WVU for implementing the project
(none had previously existed). At this point, the Mor-
gantown program was divided into three distinct
phases in lieu of the original single integrated effort to
develop a 6-station, 100-vehicle system.
  
In addition to the research and development dem-
onstration—Phase 1A—the contract between UMTA
and WVU outlined Phase IB, which would ensure an
operational 3-station, 45-vehicle system. It was also
understood that following Phase IB, WVU could ap-
ply for a capital grant (80 percent federal and 20 per-
cent local funds) to complete the 6-station, 100-
vehicle system. Included in the contract was a proviso
that if WVU were not pleased with the system, UMTA
would remove it.13
  In August 1971, UMTA selected Boeing to replace
JPL as overall system manager of the Morgantown
Demonstration Project. In a letter contract, Boeing
agreed to retain the subcontractors already selected
and to have five vehicles ready for demonstration by
October 1972. Phase 1A began.
Phase 1A of the Morgantown Project
  Following lengthy negotiations, Boeing and UMTA
concluded an agreement for Phase 1A in January
1972. As defined in the final contract, the PRT system
was to consist primarily of rubber-tired, electric-
powered vehicles with "vehicle onboard command and
control system" (VCCS) ; the guideway, with control
and communication links; stations, including station
computers; and a central-control computer. The PRT
was to run either on fixed schedules or in response to
passenger demand, at 15-second headways. Nonstop,
direct service would be provided by computerized dis-
patch and control of all vehicles. The overall com-
mand and control system would incorporate a central
computer, onboard controls, and an independent
collision-avoidance system (CAS).
  Limitations of time and money imposed by UMTA
characterized Phase 1A and virtually prevented the
use of aerospace "hardware" in the Phase 1A PRT
system. The politically motivated pre-election dead-
line, for example, required a very demanding sched-
ule. Similarly, in the interest of keeping costs down
and despite Boeing's project estimate of $44 million,
which was based on more detailed design and analysis,
the final contract for Phase 1A bore a price tag of $37
million. Several operational features of the system
(e.g., revenue service equipment, guideway heating,
operational softward, maintenance facility, etc.) were
deferred to Phase IB.
  Attempting to meet the stringent schedule within
budget fostered approaches that many considered un-
desirable in complex aerospace systems work: concur-
rent construction and design of the system's major
  
NUMBER 3

13



elements and frequent use of commercial hardware.
In September 1971, for example, ground was broken
for the guideway—before the major vehicle and con-
trol paremeters were determined.
  Furthermore, the contract stressed using "existing
state of the art technology and standard off-the-shelf
components." By definition, this implied maximum
use of traditional ground-transportation elements,
which were particularly unsophisticated because of a
long, dormant period of research and development in
the field.
  In spite of their experience and predilections, there-
fore, Boeing found they had to make some "painful"
tradeoffs in order to reconcile initial capital cost
against long-term cost effectiveness. Standard com-
mercial hardware often had to be selected in lieu of
custom-designed hardware, which would do the job
better but which might cost several times as much. For
example, UMTA encouraged less expensive brakes,
axles, and motors. Ordered through commercial con-
cerns, these parts had not undergone the rigorous
testing usually given aerospace hardware and proved
insufficiently reliable for a fully automated system.14
But the use of commercial hardward was also fostered
by the realization that if the PRT concept were to have
long-range economic viability—and thus widespread
use—it must be producible for a cost in the range of 5
to 10 dollars per pound, midway between standard
automotive technology and that of typical aerospace
applications.
  In April 1972, in preparation for the first demon-
stration, Boeing began test runs on the first five pro-
totype vehicles at its Kent test track. The "election
year hysteria" of trying to meet the October deadline
forced shortcuts and limited systems testing. It also
gave the public a false sense of progress.15
  Despite serious reservations of many responsible
participants in the program, the PRT system was
dedicated in Morgantown in October 1972. One UM-
TA official later remarked: "It was ridiculous . like
putting five 'breadboards' in the field. Anyone who
knows anything about R&D knows that what they
[Boeing] tried to do . . developing a totally new,
automated, integrated system . . couldn't be done in
the time they were given." 16
  The breakdown of one of the PRT vehicles during
the dedication was almost inevitable, and it was only a
portent. For Boeing and UMTA it was apparent that
intensive systems testing would be necessary to resolve
the problems of the prototype system design. True, the
objective of Phase 1A had been just to demonstrate the

technological feasibility of the PRT, and this had been
accomplished. But as Boeing's project manager noted,
"While no one expected this phase to result in a ma-
ture commercial system, both Boeing and UMTA were
surprised by the magnitude of the changes which
would have to be incorporated before this concept
could be translated into a public transportation sys-
tem. Then, too," he continued, "in every development
there comes a time to fly and let the real world define
the problems which remain to be solved." 17
  The testing that followed the initial demonstration
of the PRT was directed to defining those problems. It
was shaped in part by a totally unrelated event that oc-
curred during Phase 1A of the PRT: namely, an acci-
dent at San Francisco's Bay Area Rapid Transit Sys-
tem (BART), when an unmanned vehicle ran off the
end of a track.
  Because this accident on another high technology
transit system was attributed to a flaw in the auto-
mated train control system, it generated a thorough
review of the Morgantown PRT system's safety fea-
tures. During the subsequent audit and major redesign
of the PRT fail-safe systems, UMTA, Boeing, and
Bendix managers were very conscious of BART's prob-
lems. Some lessons and experience were thereby trans-
ferred from one aerospace-influenced project to
another.
  In the spring of 1973, Boeing, Bendix, and outside
experts conducted a thorough safety study. Boeing's
study was comparable to the rigorous test programs in
"fly before you buy" aircraft contracts. The test results
confirmed the inadequacy of the system's reliability
and safety, and they identified specific problems that
had to be resolved before the system would be ready
for public service.
  Most important of the safety features analyzed was
the VCCS. During the ensuing shakedown tests and
design review, major defects were uncovered in key
components of the VCCS. In the communications and
control system alone, Bendix uncovered 40 problems.
Unacceptable failure modes were discovered in the
guideway [collision avoidance systems] and in various
vehicle components such as brakes and axles.
  Ultimately, the VCCS that Boeing redesigned dur-
ing Phase IB represented the most significant transfer
of aerospace technology to the PRT's hardware. While
its individual components cannot be strictly consid-
ered aerospace technology, the specific application
and packaging of the VCCS were, according to Boe-
ing, based on criteria similar to ones important in mis-
siles or aircraft.
  
14

SMITHSONIAN STUDIES IN AIR AND SPACE



  Following the 1973 spring safety audit and its find-
ings, Boeing deepened its test program. During this
task, they approached the PRT system as they did any
complex aerospace program. For example, they con-
ducted extensive analyses to search out logic faults in
the electronic systems: "fault-tree" analyses were car-
ried out on both components and operational charac-
teristics of the system and were rigorously docu-
mented.
  In July of 1973, Phase 1A —the initial test and dem-
onstration phase—was completed. Boeing had con-
cluded that a substantially improved vehicle would be
required and that the problems defined previously
could be solved by applying aerospace rigor and disci-
plines.18
  Throughout the PRT program, the most effective
factor in the transfer of aerospace technology was un-
deniably the involvement of Boeing. During Phase 1A,
experienced aerospace engineers brought aerospace
disciplines and management techniques to bear on the
program. Proceeding according to the concepts of
"systems engineering," they defined the total system
and its performance requirements at the outset. They
then used aerospace systems engineering techniques
such as reliability and fault-tree analyses to establish
subsystem and component-design requirements.
  The technology of Phase 1B exceeded that of Phase
1A in only a few instances, e.g., in substituting redun-
dant solid state relays for gravity-actuated mechanical
relays. The big difference appeared in detailed mech-
anization of the design as a result of aerospace engi-
neering, rigor, and analytical methods. Here again,
Boeing was most instrumental in the transfer.
Phase IB
  In September 1973, UMTA allocated $20 million
and authorized Boeing to proceed with Phase IB: to
place into public service a 3-station, 45-vehicle system.
It was the first time during the program that the spe-
cific technical characteristics of the operational system
were formulated; the specifications for Phase IB were
actually the first opportunity to define the end product
on the basis of relevant tests and analysis. Previous
specifications had been stated only in terms of general
performance and design goals, many of which were
unattainable.19
  The Phase IB specifications were considered "very
sporty" for a totally new transportation system, how-
ever, and Boeing found several of them comparable to
criteria that they might have to meet in aircraft. For
example,   safety  requirements  approached   those  of

commercial aircraft, i.e., 1 X 10"4 probability of ac-
cidental fatality per operational day. This translates to
approximately one fatality per 100 million passenger
miles, less than the existing record for buses and con-
siderably less than for automobiles. Such a require-
ment was beyond the demonstrated state of the art in
surface transit systems.
  The system reliability was defined in terms of the
probability that a passenger could call a vehicle and
complete his trip within a specified time—96 times out
of 100 or better. In some instances, 10-year service-life
requirements were imposed, and operations and main-
tenance costs could not exceed $850,000 per year.
  Finally, the system had to move 1,100 passengers in
20 minutes between stations. This demanded continu-
ous dispatching of vehicles every 15 seconds.20
  During the Phase IB construction and installation
of the 45-car system, Boeing redesigned every key ele-
ment of the PRT vehicle and system. The following
elements21 are identifiable transfers of aerospace tech-
nology:
  1. A new VCCS—the "brains" of the vehicle—was
developed by Boeing "with the same degree of engi-
neering and quality control over design and produc-
tion as if it were going to the moon." The $20,000 cost
per unit is one-sixth of the cost of the vehicle, but the
VCCS is vital to the objective of 100 percent automa-
tion and thus reduces operating costs, the most expen-
sive factor in labor-dependent conventional systems.
  2. The guideway CAS is based on the concept of re-
dundancy. Two separate systems, one operated by the
computer, the other by handwired logic, check each
other continuously. If a conflict appears, the vehicles
are brought to a safe stop. A computer-controlled
management system prevents vehicles from being dis-
patched into conflicts on the guideway. In a sense this
is a third independent system.
  3. Standard railroad relays used in Phase 1A vehi-
cles for onboard switching were replaced with relays
used in spacecraft. The railroad relays were found to
be unreliable when subjected to vehicle vibrations.
The aerospace relays had high reliabiltiy and were a
fraction of the size and cost of railroad relays.
  4. A hightly responsive, redundant hydraulic brake
system was developed, using servo-valves from aircraft
applications. The original automotive-type brake
calipers were replaced with a design derived from
BART technology.
  5. Throughout, Boeing's manager of transporta-
tion systems encouraged use of more sophisticated
technology to realistically address the PRT's design
  1. 
NUMBER 3

15



requirements,  where commercial components could
not do the job adequately.
  6. Axles were redesigned and aerospace-type proc-
ess controls were imposed on the axle supplier to en-
sure a high quality product with the required system-
life.
  7. The real-time computer software developed for
Phase IB was comparable in sophistication to aero-
space programs. Industrial software used extensively
for process control and monitoring generally has a
manual backup, but the PRT program had to respond
to varying circumstances without manual intrusion.
While PRT software was not as complex as Apollo
software, it was designed to do all the same things on a
smaller scale.
  8. Boeing used an aircraft computer program to
analyze the maintenance requirements of the PRT.
  9. Military and aircraft standards were specified for
selecting components. As in aircraft, certain compo-
nents were designated "safe-life designed," and their
mandatory use was specified in the maintenance man-
ual. The manual itself resembles those used by Boeing
for aircraft or missile systems.
  10.	Self-diagnosis, like that found in aerospace sys-
tems, was built into Phase IB vehicles. Vehicle faults
were relayed at once to the central computer. Redun-
dant systems could propel the vehicles to repair facili-
ties if problems were found in the primary systems.
  As a prerequisite to production of the Phase IB ve-
hicles, Boeing had reviewed with UMTA all fault-tree
analyses and hazard reports, focusing attention on
safety and reliability. UMTA then had conducted a
thorough review of the design and had awarded Boe-
ing a contract for the first phase of production. For
this task, Boeing also brought its aerospace
experience22 into play:
  1. The manufacturing manager was transferred
from the Boeing 747 aircraft to the PRT vehicle. In his
new role, he used virtually the same techniques of pro-
duction scheduling, crew loading, budgeting, and
plotting of learning curves that were used for aircraft
and missiles.
  2. The manufacturing facility in which the vehicles
were assembled was the same used to build the super-
sonic transport (SST) mockup; the vehicles shared
the building with another prototype airplane.
  3. Engineering quality analyses, similar to those
performed in aerospace work, were performed on key
PRT vehicle components. For example, as items ar-
rived from the supplier, assemblies were dismantled
and tested for tolerances and workmanship.
  
By September 1975, Boeing had extensively tested
and debugged all 45 vehicles at the Seattle test track
and in Morgantown. Throughout this task they relied
on "highly systematic" approaches (the use of formal
test requirements, plans, and analysis reports) that are
not exclusive to aerospace but are more common to
that discipline than to mass-transit.23
  Following the tests, UMTA accepted delivery of the
system, and Boeing signed a contract with WVU for
assistance and training during the first year of opera-
tions. UMTA also approved $4 million in capital
grants for additional facilities, first year operating
costs, and preliminary design for a Phase II expansion,
whose approval was pending as of the end of 1976.
  This marked the formal completion of the demon-
stration phases of the Morgantown PRT project. Ac-
cording to the agency's Morgantown program man-
ager, Boeing's performance was one of the brightest
spots in the program. He further considered that the
aerospace input to Phase IB —especially systems ap-
proaches and discipline— had saved the project.24
  Before the Morgantown demonstration project was
completed, UMTA had initiated a so-called "High
Performance" PRT demonstration. Lessons learned in
Morgantown may be transferred to this second-gener-
ation PRT most directly through the participation of
Boeing, which won one of the three contracts awarded
by UMTA for preliminary designs.
Potential for Future Use of PRT Technology
  Many aerospace companies (e.g., Boeing, Rohr,
Westinghouse, LTV, Bendix) have made major in-
vestments in the mass-transit field. It was a natural
market for them, and national emphasis on solving ur-
ban transportation problems sustains hope that it will
eventually be a large, dynamic market.
  The developers of new transit systems, however,
have had serious difficulty in finding and convincing
potential customers: transit operators, private in-
vestors, city planners, or consultants to the city. In
general, the problem stems from several factors. First,
transit operators ^re frequently preoccupied with the
problems of existing capital facilities. Their economic
concerns are described by their constant search for
operating subsidies. Advanced technology is beyond
their interest or finances.
  Second, city planners are not well acquainted with
the potential of new transit systems. They instinctively
choose systems with the lowest initial acquisition cost
rather than with the lowest life-cycle cost, the benefits
  
16

SMITHSONIAN STUDIES IN AIR AND SPACE



of which accrue only in future years when they are
likely to be out of office.
  Finally, communities choose one system over an-
other on the basis of consultants' recommendations.
Reflecting the interests of their political mentors, con-
sultants may be conservative and reluctant to recom-
mend what is not operating and available. They tend,
therefore, to steer their clients away from the R&D
business with its attendant risks.
  Boeing and other aerospace contractors have tried
to market total systems on a "turnkey basis," that is,
selling to communities a completed transportation
"package," ready for operations. From its experience,
Boeing concluded that buyers aren't ready for this ap-
proach to innovative design and installation. The
customer seems to care little about technology and
management processes, focusing instead on his imme-
diate needs: Do I get there when and how I want and
for what I want to pay? Such questions are easier to
discuss in terms of existing products rather than future
systems with unmeasured benefits. It is likely that the
market for such systems will develop only after the
system supplier and the buyer can agree on language
of long-range benefits.
  In the interim, Boeing's bid on the high-perform-
ance PRT reflects optimism that PRT's are a part of
the future of transit and that the government will con-
tinue to support this technology.
  In fact, the future of the PRT would seem to de-
pend on several significant factors, only one of which
involves the technology itself: first, successful opera-
tion at Morgantown; second, public acceptability of
the characteristics of automated systems; third, more
progressive city planners; fourth, federal policy per-
mitting PRT eligibility for capital grants; and finally
the extent to which automobiles are excluded from the
inner city.
Critical Assessments of the Morgantown PRT
  As suggested throughout this study the major targets
of criticism directed at the Morgantown PRT were its
cost, complexity, and technical problems—especially
when compared to its limited use in a small-town en-
vironment. At first glance, the criticisms seem valid.
At $64 million, the Morgantown six-mile system, serv-
ing a community of 40,000, is expensive by any stand-
ard. The case against it is even stronger in the face of
critical, unsolved transit needs of cities 10 to 30 times
more populous. Unquestionably, the PRT is also com-
plex and relies heavily on highly sophisticated com-
ponents.
  
Nevertheless, there are several valid counterpoints
that are often obscured by the controversy and ignored
by avid critics of the PRT project.
  First, the Morgantown PRT was a "demonstration"
project that attempted to introduce a new solution to
one of the nation's many transit problems: the con-
gested inner city or small urban area. No economically
viable approach had been advanced to date. The PRT
was conceived not only as one community's transit sys-
tem, but also as an operating test-bed to provide em-
pirical data on production cost, performance, reliabil-
ity, and public acceptance.
  These are legitimate functions of a demonstration
project, which, in turn, is an effective means whereby
the government may transfer technology to the other
markets. The government correctly assumes that role
where the cost of the technology would preclude its in-
troduction by industry alone. Measured by these crite-
ria, the Morgantown PRT represents a significant
achievement: a totally new application of advanced
technology to an operating urban system, which may
provide a cost-effective solution to mass transit in the
future.
  Second, the Morgantown PRT was typical of devel-
opment projects, for which the costs and technical out-
come are unpredictable. It was inherently subject to
costly and extensive testing and redesign. In fact,
UMTA made no great claims of predictability to Con-
gress. Less than six months after approval of the
Morgantown demonstration project, Robert Hemmes,
assistant administrator for program demonstrations,
said in appropriations hearings of the 1972 fiscal year:
  Our estimates began at $1 million a mile. But as more was learned
about these systems, they have followed the same pattern of cost
escalation that all high technology projects have followed. That is,
at first there are many considerations that are not in the estimate
because we haven't encountered those problems yet. Later the esti-
mate grows as we build the system and we encounter the problems.
  The $1 million a mile is for a system without switching and merg
ing. When we added the switching and merging capabilities, the
estimates were $5 million a mile. Now that we have encountered
problems, which was to be expected, and are following the familiar
pattern of technology development in other past programs such as
aerospace and defense programs, the estimates have grown to $7
million and in some of the more complicated ones to $9 million a
mile. But if we carry 7,000 people per hour past a point at $9 million
a mile, this would compare favorably with the construction of an
eight-lane highway carrying the same number of people. The high-
way, of course, isn't very complicated in technology but it requires so
much right-of-way in congested urban environments that it could be
much more expensive than people-movers.8S
Of course, the "$9 million a mile" also proved gross-

NUMBER 3

17



ly underestimated. But to those critical of the demon-
stration's $64-million price tag, UMTA officials main-
tained that it was entirely justifiable and comparable
to any research and development program as innova-
tive as the PRT; its effective 18 percent cost overrun
was actually far less than many complex R&D pro-
grams.
  To put the Morgantown PRT in perspective, its cost
should be compared to the development costs of other
such projects or competitive mass-transit systems.
Although the PRT has been expensive to build, its
complete automation reduces its operating costs
(while permitting it to operate on 24-hour demand),
thus making it far less expensive over the long term
than a bus system of the same capacity and service
capability. (The break-even point between the two
modes is estimated at about seven years.)
  Nevertheless, UMTA officials realized that the high
initial cost of the PRT would militate against its
widespread replication. Now that the technology has
been virtually proven, UMTA expects that the next
generation system will carry lower cost of replication.
  Despite these justifications, however, it was the pro-
gram's mounting cost and its unenviable distinction as
one of the banes of UMTA's relationship with Con-
gress that made it something of a "programma non
grata" in DOT. The Congressional Appropriations
Committee became increasingly critical of PRT's cost
overruns. The issue was further confused by the
bizarre spectacle of WVU threatening to invoke its
contract and compel DOT to remove the whole system
(at an additional cost of $7 million) if the agency did
not fund completion of the system to its original
scope.26
  Of course, the problem of convincing Congress to
fund development projects was not unique to UMTA.
Backers of the SST bear witness. It also illustrates the
critical relationship between such requests and current
national priorities. Congress is more easily convinced
when a request seems to support a stated national
goal; it is less easily swayed when other goals take
precedence or funding is tight. At the same time, the
1964 legislation creating UMTA provided for "re-
search, development, and demonstration projects in
all phases of urban mass transportation." The agency
had a mandate to conduct such programs, and Con-

gress had a responsibility to understand what they en-
tailed.
  A third factor which should be kept in mind is that
the system's complexity was a function of its competi-
tion with the automobile for public acceptance. At the
least, the system's acceptability depended on maxi-
mum safety and reliability, maximum convenience
and comfort, and minimum operating costs. To
achieve all this within a specified number of dollars
and level of complexity was a major challenge.
  Fourth, a complex R&D program cannot be subju-
gated to the vagaries of internal politics without
substantial adverse impact. When politics enter into
technical decisions, long-term benefits are almost in-
evitably sacrificed— at great cost— to short-term gains.
Involved in the process, the PRT's technical people
resented the politically inspired decisions of the
administration, which adversely affected the pro-
gram.27
  On another level, past government experience with
complex R&D programs had also proved that sound
management is the key to success. An agency lacking
such expertise, therefore, should not assume the role
of systems manager. In the case of the PRT, stresses
within the program were reflected—and intensified —
by the management weaknesses in the Federal Govern-
ment. During the first three years, for example, the
Morgantown PRT had several project managers, was
shifted between two administrators, and was a major
cause of the resignation of an assistant administrator.
  Finally, in evaluating the Morgantown project, it
should be remembered that the PRT was an at-
tempt—one of few—to get the auto out of the city
center. It may not be the optimum solution, but in the
absence of any viable alternatives, it deserves fair trial
and evaluation. Although it is initially expensive, its
automation may ultimately yield a more favorable cost
picture over its entire life. Boeing's Morgantown PRT
project manager noted: "The problems encountered
at Morgantown in this contract are lessons learned and
were worth the money they cost; there was no other
way to gain the experience. The tragedy of Morgan-
town will be if it suffers a premature demise brought
about by political considerations and critics who do
not know how to run or evaluate a large scale tech-
nology experiment.'' 28
  
18

SMITHSONIAN STUDIES IN AIR AND SPACE

Notes

  1. Interview, Dr. Samy E. G. Elias, chairman, Industrial Engi-
neering Department, West Virginia University, 19 November 1974.
2. Ibid.
3. Ibid.
  4. Ibid.; and " 'Morgantown's People Mover': A Little Engine
That Couldn't," Washington Post (2 June 1974)   Section C, page 3.
  5. Elias interview; J. P. Cunliffe, Frederic R. Harris, Inc., "Per-
sonal Rapid Transit," paper presented to the Society of Automotive
Engineers, Earthmoving Industry Conference, 2-4 April 1973. Soci-
ety of Automotive Engineers, Inc., publication number 730439
(New York, 1973) ; House of Respresentatives, hearings before a
subcommittee of the Committee on Appropriations, 92nd Congress,
1st Session, "DOT and Related Agencies Appropriations for 1972,"
part 3, pages 242-245, 454-457; interview, Dr. George Pastor, as-
sistant adminsitrator for R&D, Urban Mass Transportation Admin-
istration, DOT, 17 October 1974.
  6. Elias interview; " 'Morgantown's People Mover' "; Pastor in-
terview.
  7. House of Representatives, hearings before a subcommittee of
the Committee on Appropriations, 93rd Congress, 2nd Session,
"DOT and Related Agencies Appropriations for 1975,' part 4,
pages480-483, 569-571, 680-683,722-737.
  8. Interviews with these Boeing Aerospace Co. management per-
sonnel at the Kent Space Center, Seattle, Washington, 14 October
1974: T. M. (Scotty) Davidson, Morgantown project manager;
John W. Allen, Morgantown systems integration and test manager;
Jack H. Keeney, Morgantown product support manager; and
Harvey Iverson, Morgantown manufacturing manager.
9. Ibid.

10. Elias interview.
11. Pastor interview.
12. Elias interview; Pastor interview.
  13. Elias interview; Pastor interview; "The Escalating Cost of a
People Mover," Business Week (16 March 1974), pages 50-52.
  14. Letter from T. M. (Scotty) Davidson, Boeing Aerospace Co.,
1 October 1975.


  15. "The Escalating Cost"; John K. Geiger and William L. Hess,
"Electrical Features of the Morgantown Project Vehicles, Power
Supply, and Control Systems," IEEE Transactions on Industry Ap-
plications 1A-9, number 2 (March/April 1973), pages 248-255;
interviews with Boeing Aerospace managers.
  16. Pastor interview. The UMTA official was also speaking from
his own extensive experience in the aerospace business and as head
of a new department in the Transportation Systems Center (former-
ly under NASA), whose early assignment was to provide technical
support to UMTA for Morgantown.
17. Davidson letter.
  18. Interviews with Boeing Aerospace managers; Davidson let-
ter.
19. Interviews with Boeing Aerospace managers.
20. Ibid.; Davidson letter.
21. Ibid.
22. Interviews with Boeing Aerospace managers.
23. Davidson letter.
  24. Interview, Steven A. Barsony, director, Morgantown Divi-
sion, UMTA, 21 January 1975.
  25. House of Representatives hearings, "DOT and Related Agen-
cies Appropriations, 1972."
  26. The university subsequently accepted the Phase IB system,
signed a contract with Boeing for support services during the first
year of operation, and looked forward to possible capital grants to
extend the system.
  27. It should be noted that they felt equally frustrated by a "cyni-
cal and headline-seeking press," which they believe accentuated the
negative.
28. Davidson letter.
  Additional Source: United States Congress, Office of Technology
Assessment, Automated Guideway Transit: An Assessment of PRT
and Other New Systems (Washington, D.C..June 1975).
  
Liquefied Natural Gas Tankers
Susan Frutkin

EDITORS COMMENT
  The second of Susan Frutkin's studies is timely in its connection to
U.S. energy problems. Building on a foundation of research and
economic analysis that Booz Allen Applied Research performed in
1973, she conducted numerous interviews and then refined and ex-
tended the story to describe processes involved in the transfer of
aerospace technology to liquefied natural gas tankers. While she has
emphasized the importance of individuals and organizations in de-
veloping liquefied natural gas (LNG) technology, she is no less care-
ful to present relevant technical material.
  Frutkin has made several points in explaining why aerospace
technology was used on LNG tankers. The techniques of storing and
handling cryogenic substances, developed by aerospace contractors
in the last two decades, were readily adaptable to use at sea. Many of
the safety and control devices for handling liquid gas were taken
from the missile and space industry. The fact that General Dynam-
ics, which owns Quincy Shipyards where many LNG tankers are
built, is a large aerospace contractor induced them to use their aero-
space engineers. Much of the technology of potential use, it turned
out, was too sophisticated or was unsuitable. Frutkin implies that
some resistance developed because space technology was imposed on
nautical architects and engineers with different traditions and orien-
tation. In the low-temperature containment system, 9 percent nickel
steel was replaced by an aluminum alloy, apparently because it was
more familiar to the aerospace engineers. Fracture mechanics, a
subdiscipline of the mechanics of materials used frequently in aero-
space industry, was applied to test ship structures in advance. The
method helped to certify the safety of the ships and thus to gain rap-
id approval from federal agencies, which were charged with con-
firming guarantees and subsidies. Aerospace technology did not
bear the blame, but its reputation may suffer from a 1973 explosion
of an LNG storage tank on Staten Island, which killed some 40 peo-
ple. If the LNG program is deemed unsafe, there will be no gain
from the cancelled aerospace technology within it.
  Frutkin's major point is that space technology was transferred to
LNG tankers because management insisted. The corporate commit-
ment of General Dynamics to aerospace methods moved people
from its Convair Division to Quincy, from spacecraft to ships. That
commitment overcame traditional differences between workers from
Quincy and Convair and permitted more or less harmonious rela-
tions while producing the tankers. The arrangement fashioned a
unique aerospace-nautical project.
Profit is still the definition of success in the U.S. capitalistic econ-
Susan Frutkin, formerly with Booz Allen Applied Research, Bethes-
da, Maryland, now with American Petroleum Institute, Washing-
ton, DC. 20037.

omy. Could LNG ships have been built profitably at Quincy without
aerospace contributions? We know at least that they could have been
built, because LNG tankers have been sailing under foreign flags for
almost two decades. Yet many results of tapping General Dynamics'
aerospace expertise contributed to the safety and utility of these
large ships. We fear that one cannot separate the aerospace compo-
nent of economics from the rest of the profit (or loss) of LNG tank-
ers. We will know if the project has been successful economically
only in future years. If it is, the energy crisis in America will have
played a major role, but space technics will share a measure of the
success.
Introduction
  In the late 1960s, several factors combined to stimu-
late U.S. interest in LNG tanker development. A rapid
decline in U.S. shipbuilding, an unfavorable balance
of payments, and a strong domestic cryogenic capabil-
ity coincided with decreasing energy supplies and the
thrust for a cleaner environment.
  As awareness of an impending "energy crisis" grew
in the sixties, abundant and clean natural gas became
an obvious candidate for petroleum substitution. Until
then, U.S. gas supplies and reserves were sufficient to
meet existing domestic demand. After the mid-1960s,
however, energy consumption rose, while exploration
and development of U.S. natural gas resources were
discouraged by price ceilings imposed by the Federal
Power Commission (FPC) in 1954. Although Alaskan
fields were opened, importation to the lower 48 states
was inhibited by the expense of transporting gas and
by a provision of the Jones Act that requires transport
in unsubsidized U.S. flagships, of which none designed
to carry natural gas were yet in service. By 1970, U.S.
demand for natural gas exceeded domestic supply by
five percent, and the Federal Power Commission esti-
mated the deficit would reach approximately 40 per-
cent by 1990.'
  The main sources of developed natural gas reserves
are Algeria, Australia, Borneo, Indonesia, Iran,
Libya, Malaysia, Nigeria, Trinidad, USSR, and Vene-
zuela. With one exception, all are separated by oceans
  
19

20

SMITHSONIAN STUDIES IN AIR AND SPACE



from the United States and the other two major natu-
ral gas markets: Europe and Japan. By definition,
therefore, increased demand for natural gas stimu-
lated import programs and increased requirements for
ships to transport the gas by sea.
  Because the ratio of the volume of natural gas at
sea-level pressure and room temperature to liquid
natural gas is 618 to 1, the viable economic way to
transport the gas is in its liquid state. It is liquefied by
cooling to its boiling temperature of about —258° F;
Thus LNG is characterized as a "cryogenic" substance.
Taken together, the properties of LNG produce a
maze of technical problems in handling, which arise
from extreme temperatures, high volatility, trouble-
some fluidity, metallurgical sensitivity of containers,
hazardous chemical properties, and rigid require-
ments for transfer and conversion of LNG. These
characteristics impose design criteria on a ship to a
degree not approached by any other commercial car-
go.2 They make LNG tankers not only technologically
unique but also more expensive than any other cargo
ship of or near their size.
Technological Requirements of the LNG Tanker
  During ocean transport, LNG must be maintained
as a saturated liquid, i.e., at cryogenic temperatures
and near atmospheric pressure or close to the point at
which it turns to vapor. Because cryogenic tempera-
tures are far below the brittle fracture point of or-
dinary ship steel, contact with the hull risks serious
damage and even imperils the ship in heavy seas. In
addition, gas leaks must be carefully controlled to
avoid potentially explosive conditions.
  The critical elements of an LNG tanker, therefore,
are the cryogenic systems, of which the containment
system is the major element. The containment system
is principally a holding tank or tanks, generally de-
scribed by its components and configuration: primary
and secondary barriers, high performance insulation,
instrumentation, shape, and interfaces with the ship's
hull. Since the first work was started on these ships,
several different versions have been developed that
comprise two generic designs of the containment sys-
tems: (1) free-standing tanks—prismatic, cylindrical,
or spherical versions of self-supporting tanks utilizing
various barrier materials and insulation but all struc-
turally independent of the ship's hull; (2) membrane
tanks—thin layers fitted into a tank composed of the
ship's hull and load-bearing insulation.3
Safety and control systems are as important as the

containment system. LNG tankers typically contain
sophisticated leak-detection devices, fail-safe systems,
pressure and temperature sensing and control instru-
mentation, vent and pressure-relief systems, and boil-
off vapor disposal (often integrated into the propul-
sion system as boiler fuel), in addition to ordinary fire
and safety equipment. Handling and storing LNG in-
volves other special systems and equipment for making
tanks inert; drying, cooling, warming, and aerating
tanks; pumping, loading, discharging, and measuring
cargo.
  While not related specifically to the cryogenic sys-
tems, the relatively low density of LNG and the
containment system give the ship an unusually high
freeboard. Attendant problems of stability must be
counteracted by supplementing ballast, for which
space is limited, with special stabilizers.
  Finally, all elements of the tanker that come in con-
tact with the LNG must be compatible with low tem-
perature conditions. They must be designed to mini-
mize thermal dimensional changes and stresses, and
they must be durable in the marine environment as
well.'* The net result is a unique, single-purpose ship,
whose primary design criteria are safety-oriented, and
whose price exceeds $100 million (for a 125,000-
cubic-meter vessel).
The Context of Operations
  A typical project for importing natural gas to the
United States requires a comprehensive system of
which the tanker is only one element. The first phase
of the system involves extraction, field processing,
pipeline transportation, compression and liquefaction
of natural gas at facilities near the source of supply.
The marine transportation phase brings LNG in tank-
ers to receiving terminals at a U.S. port (usually near
highly populated market areas). There, at shoreside
terminals, the LNG is unloaded, transferred, stored,
and revaporized to natural gas for distribution to con-
sumers (who may not be in the same state).
  This comprehensive project represents a significant
financial commitment. The costs of implementing one
Algerian-U.S. project for importing 1 billion cubic
feet of LNG per day over 25 years included invest-
ments in shoreside facilities (gas wells, gas-gathering
system, processing plant, pipeline, liquefaction plant,
storage and loading facilities) and were estimated to
exceed $1.1 billion; the additional cost of the requisite
nine LNG tankers exceeded $800 million.5
In addition to the costs, the complexities and diffi-

NUMBER 3

21



culties in bringing about such a project have been apt-
ly compared to the "Perils of Pauline." 6 Involved are
coordination, agreement, and approval of a stagger-
ing array of organizations and individuals concerned
with investments, safety hazards, or political and eco-
nomic implications of importing LNG.
  At the outset, gas supplier (company or govern-
ment) , shipowner, shipbuilder, ship lessee (which can
be many different people or institutions), gas storage
companies, distributors, utilities, bankers, and insur-
ance companies must work together. To implement
the project and certify the ships also necessitates meet-
ing standards, filing applications, and gaining ap-
proval of a multitude of public authorities. These in-
clude the American Bureau of Shipping (ABS), U.S.
Coast Guard (USCG), Federal Power Commission
(FPC), Environmental Protection Agency (EPA),
Commerce Department (Bureau of Customs, Meas-
urement of Vessels), Maritime Administration
(Mar Ad), Federal Communications Commission
(FCC), International Telecommunications Commis-
sion, Army Corps of Engineers, Department of Inter-
ior, State Department, Department of Transporta-
tion, International Convention for the Safety of Life at
Sea, Intergovernmental Maritime Consultative Organ-
ization, state and local governments and their agencies
(including port authorities, energy and environmental
control offices).
  Complicating this process still further, public agen-
cies often hold hearings, thus involving a multitude of
private and public interest groups.
  In all, over 60 professional and governmental enti-
ties, plus private interests, could be involved in a single
project to import natural gas into the United States.7
Potential for Transferring Aerospace Technology to
LNG Tankers
  The element of the LNG tanker most suited for
adoption of aerospace technology is the cryogenic con-
tainment system. The big space and missile programs
of the 1960s considerably advanced technique in cryo-
genic engineering. It should be noted that much of the
cryogenic work conducted for the space program in-
volved improvements in technology that originated in
traditional cryogenic industries, i.e., chemicals, ener-
gy, food, and metallurgy. The designation "aerospace
technology" is therefore applied with some caution.
  Significant space-inspired advances from handling
cryogenic substances were made in insulation, metal-
lurgy,  adhesive welding, instrumentation (e.g., cus-

tody transfer system developed for Apollo spacecraft to
measure and test liquid fuel taken on), analytical
methods (stress analysis, fracture mechanics, thermal
dynamics), and systems management.8
  In the important area of insulation, several aero-
space suppliers developed significant innovations.
Notable were the Martin Marietta Company (internal
honeycomb insulation using a bleed and vaporize con-
cept to create a pressure barrier) ; Rockwell Interna-
tional (wet-wall urethane insulation, now licensed to
French and Japanese concerns), and McDonnell
Douglas Astronautics Company (three-dimensional
fiberglass matrices, reinforced urethane, internal insu-
lation systems) .9 All were developed in conjunction
with the space program, and all have potential LNG
tanker application; however, none has yet been ap-
proved by the U.S. Coast Guard for use in ships.
  Each of the organizations and individuals that are
involved in coordinating, financing, and approving a
natural gas import project is a potential "inhibitor" or
"promoter" of technology transfer. Because they are
most often involved on account of the considerable fi-
nancial commitments or safety hazards, they tend to
approach technological innovations conservatively.
Those who finance such projects place a high pre-
mium on maximizing returns and minimizing costs
and risks. Those concerned with safety seek assurance
through time-tested guarantees of performance. Eco-
nomics and safety may therefore be decisive factors in
transferring aerospace technology to the LNG tanker.
  At the same time, political and social problems,
delays, uncertainties, long lead times, and long-term
advance commitments may greatly influence the po-
tential for transferring technology, and they may ul-
timately outweigh both technical merits and other
promoters of transfer.
  Another group of potential transferors is less direct-
ly affected by these influences. Subsystem contractors,
material and hardware suppliers, ship designers, and
patent holders are all participants in the LNG tanker
development but are one step removed from the finan-
cial and regulatory institutions. Nevertheless, to the
degree that they are subject to the shipbuilders' speci-
fication and wish to do business in U.S. yards, they too
are subject to the same constraints.
Evolution of LNG Tankers
  As noted, the distinguishing feature of LNG tankers
is the cryogenic containment system. The evolution of
these ships is therefore reflected in its changes.
  
22

SMITHSONIAN STUDIES IN AIR AND SPACE



  The earliest development of LNG tankers was done
in the United States. In 1952, energy and shipbuilding
interests experimented with balsa-lined, free-standing
tanks fitted onto a 6300-cubic-meter barge to transfer
LNG from Louisiana to the Chicago stockyards. This
first attempt to transport LNG by water was a relative
failure; tests showed surface damage to the wood insu-
lation resulting from direct contact with the LNG.
  Between 1957 and 1959, an American gas company,
a shipbuilder, and a naval architect jointly ventured
with the British Gas Board to prove the feasibility of
large-scale ocean transport of LNG. In this experi-
ment, a 5000-cubic-meter dry cargo ship was fitted
with five aluminum alloy balsa-polyurethane insula-
tion freestanding tanks. This ship carried the first
oceangoing LNG from Louisiana to England. The ex-
periment was terminated in 1961 after six successful
voyages.
  For more than a decade following these early proj-
ects based in the United States, virtually all design and
development of LNG tankers was conducted in
Europe. Lack of U.S. involvement stemmed largely
from the lack of need or incentive to import foreign
gas. At the same time, the greater demand in Europe
and French interest in exploiting Algerian resources
stimulated development of LNG tankers abroad.
  During this period, European energy interests made
several refinements, most significant of which were in
insulation materials, metal alloys, tank shape, barrier
construction, utilization of boil-off as auxiliary fuel,
and submerged cryogenic pumps. European interests
patented a number of LNG containment designs, on
which were based all LNG tankers built or under con-
struction through 1975. Furthermore, until 1972, all
18 LNG tankers ordered worldwide were either deliv-
ered from or under construction in European yards.10
  There are indications that several innovations uti-
lized in the European tanker designs reflected transfer
of U.S. aerospace technology abroad.11 While the full
extent of this transfer has not been documented, some
of the major concepts, materials, and processes can be
traced to the United States. Among these, for exam-
ple, was a submerged cryogenic pump, considered a
major step forward in LNG cargo-handling technol-
ogy. The pumps were placed in two ships that were
part of an all-European energy project and had been
adapted by a U.S. supplier from similar designs used
in aircraft.12 Much of the hardware for the ships was
procured in the United States.
  Other examples of transfer of U.S. technology to the
European   designs   are   evident   in   the   Norwegian

Kvaerner-Moss spherical tank design, later licensed to
General Dynamics for production in the United States.
The tank support ("ring equator") is comparable to
that used for the Saturn V oxidizer tanks. In addition,
metal alloys, welding techniques, and insulation mate-
rials used in the ships were all supplied by U.S. com-
panies.13
  It was not until 1972 that the first order for LNG
tankers was placed in the United States. Since then, a
total of 16 ships has been ordered in four U.S. yards.
While all are being built under license to European
patent holders, the potential for transferring U.S.
aerospace technology existed nonetheless. To see how
such transfer may have been achieved, Booz Allen
Applied Research selected the case of Quincy Ship-
yard's LNG tanker program for in-depth research and
analysis. Quincy was chosen because (1) it was the
first U.S. yard to conclude a contract for LNG tank-
ers and is constructing half of the tankers on order in
the United States now, and (2) it is a wholly owned
subsidiary of General Dynamics, a major aerospace
conglomerate. It therefore presented an excellent op-
portunity to study interaction between the aerospace
and marine industries.I4
Transfer of Aerospace Technology to the Quincy
Shipyard and to General Dynamics' LNG Tanker
  General Dynamics acquired Quincy Shipyard in
1964.15 A steep decline in ship orders over the next
several years culminated in the loss of several large
Navy contracts for which Quincy had been a prime
contender. Work on the shipyard's current orders was
to be completed by 1972, and General Dynamics con-
sidered closing the yard after that. (Several thousand
workers were actually laid off before work on LNG
tankers was eventually begun.) Among alternatives for
revitalizing the yard was the possibility, pursued by
both Quincy and General Dynamics' corporate man-
agement, of entering the LNG tanker market to ex-
ploit the predicted rise in international energy trade.
  Simultaneously, representatives of European patent
holders were trying to market their designs worldwide.
In the case of the Moss-Rosenberg shipping group,
holder of the patent on the Kvaerner-Moss LNG
spherical tank system, their small yard was limited to
building one ship at a time.
  In spite of the obvious potential, however, the con-
siderable financial investment and risk in the new ven-
ture prompted General Dynamics to undertake two
years of study and evaluation before making a com-
  
NUMBER 3

23



mitment and signing a license agreement.
  The first phase of the study included a thorough ap-
praisal of all recognized LNG containment systems.
This review was made under the direction of the gen-
eral manager of Quincy Division and in cooperation
with the General Dynamics corporate vice president of
Engineering and Program Development. Thereupon
Quincy selected the Moss-Rosenberg spherical tank
system as most readily adaptable to the yard's capabili-
ties. The decision was based principally on the fact
that alternative, membrane systems were labor inten-
sive, a distinct disadvantage to U.S. yards, where labor
costs were not competitive with foreign yards. By con-
trast, the Moss-Rosenberg tank was not only less
labor-intensive, but it was also considered more pre-
dictable in operation and easier to maintain.
  In mid-1970, Quincy Division proposed to the Gen-
eral Dynamics Board of Directors that the company
enter the LNG tanker market. Further review of the
containment system design and a Det Norske Veritas
analysis culminated in the signing by General Dy-
namics of an exclusive licensing agreement with Moss-
Rosenberg.
  For still another year, Quincy management con-
ducted extensive studies to support their request for
corporate financial backing necessary to enter actively
the LNG tanker market. A market strategy study indi-
cated that LNG tankers were a good long-term busi-
ness opportunity and had the highest commercial
value per square foot of any kind of ship under consid-
eration. Manufacturability and facility analyses were
also conducted by several shipyard engineers, working
with an engineer who had experience both in the
Douglas Aircraft DC-9 program and in Litton Indus-
tries' automated shipyard at Erie.
  Finally, structural risk analyses—which proved to be
a crucial determinant—were performed on the design
of the containment system. An interdisciplinary team
was formed for this effort, which included naval engi-
neers and aerospace engineers with experience in
cryogenic structures and metal fracture mechanics.
The latter were "borrowed" from General Dynamics
Convair Division, where the group's leader had par-
ticipated in the Atlas-Centaur program. The team
concluded that the Moss-Rosenberg design was well
"within the state of art" and could use off-the-shelf
hardware in "everyday use." This was the first of many
instances duing the company's LNG tanker produc-
tion program in which the Convair Division was called
upon for assistance and in which analytical methods
were used effectively.
  
In spite of the team's convincing arguments, how-
ever, corporate approval might still have been with-
held had Quincy's efforts not coincided with Federal
Government incentives. The Merchant Marine Act of
1970 made bulk-carrying vessels eligible for the first
time for construction and operating subsidies and con-
struction loans and mortgage insurance. In addition,
the Internal Revenue Service authorized investment
tax credits for the first time for owners of tankers in in-
ternational trade. Devaluation of the dollar in 1971
was an additional mitigating factor.
  Applying these advantages to LNG tankers was not
automatic, however. LNG tanker subsidies, for exam-
ple, required MarAd approval, which was slowed by
the cost of LNG tankers, considerably higher than
cargo ships of the same size. Application of the 45 per-
cent construction differential subsidy (CDS) to a
number of LNG tankers costing almost $100 million
each was clearly beyond MarAd's budget. Quincy ex-
ecutives worked closely with MarAd administrators
(whom they characterized as "creative and entrepre-
neurial bureaucrats") to arrive at an arrangement
whereby LNG tankers would be subsidized to the level
of the most expensive conventional ship that already
qualified. Ultimately, General Dynamics received a
23.7 percent CDS on its first three LNG ships.16
  Although less significant as a direct economic incen-
tive, a three-phase support program was initiated by
MarAd in 1971 to reinforce tanker subsidies and to
further facilitate all LNG import projects. At least one
rationale behind the effort was the desire "to capitalize
on this nation's know-how in cryogenics based on space
and missile program experience. The U.S. aerospace
industry has extensive technological capability which
is not now being applied to Marine systems."17 MarAd
officials calculated that, because of the highly devel-
oped U.S. cryogenic support industries, with technol-
ogy that could be extended to design, fabrication, and
installation of shipbome cryogenic systems, "it would
appear that certain LNG cryogenic containment sys-
tems can be more competitively built in the United
States than elsewhere in the world, provided that
standards and continuity of programs are established
and maintained." I8
  The first phase of MarAd's program consisted of
testing and refining (European) containment systems.
Eventually projects were initiated that included a liq-
uid sloshing measurement, tank design, materials test-
ing, leak detection, and design of instrumentation for
custody transfer, the last at General Dynamics' re-
quest.
  
24

SMITHSONIAN STUDIES IN AIR AND SPACE



  The second phase of MarAd's program was a short-
term effort to reduce the economic risks inherent in
LNG shipbuilding. An economic risk analysis was con-
ducted, the results of which predicted a decline in de-
mand for LNG tankers by 1985-1990. Shipbuilders
were also encouraged to identify high risks and high
costs that could be reduced through a shared R&D
program.
  The third phase was to have been a long-range R&D
effort directed to reducing costs and ameliorating risk
in second generation containment systems through the
use of U.S. (aerospace) cryogenic technology. This
phase was never fully implemented due to the pessi-
mistic conclusions of the economic risk analysis. (Sub-
sequently, these conclusions were reinforced by a shift
in national priorities from clean energy to energy self-
sufficiency.) Nevertheless, MarAd initiated several
R&D cost-sharing projects with companies already
working on advanced cryogenics developments (two of
which were Rockwell International and Martin
Marietta, whose insulation designs were noted above).
MarAd also supported a small program for trans-
ferring experimental cryogenic technology to ship-
builders (essentially information dissemination) from
the Cryogenics Division of the National Bureau of
Standards.19
  In mid-1971, the General Dynamics board of direc-
tors approved entrance into the LNG tanker market
and committed roughly $40 million20 in capital in-
vestments to Quincy Shipyard Division. Two 900-foot
by 150-foot basins and an 850-ton lift crane to install
containment tanks in the ship accounted for most of
the funds expended in the LNG tanker program.
  Following General Dynamics' decision to enter the
LNG tanker market, the containment system design
was subjected to rigorous fracture-mechanics analyses
(based on principles developed during the company's
work to resolve F-lll problems), and refinements
were made in design criteria. According to Quincy's
chief development engineer in charge of performing
the stress analysis, "this whole design procedure was
unfamiliar in shipbuilding. Without the aerospace in-
dustry, fracture mechanics would not have been de-
veloped to the point of anyone thinking of invoking it
as a design criterion for ships. It was beyond the state
of the art in ships, but not in aircraft." 21
  The analysis assured a structurally sound ship, a
conclusion that was of value because General Dynam-
ics' 125,000-cubic-meter LNG tankers were larger
than any previously built.
  
The frequent resistance to transferring technology,
even within a corporation, is illustrated by the obser-
vation of one of the participants that Quincy was
"forced" to seek and use the advice of its sister divisions
(Convair and Electronics) .22 The shipyard (tradition-
ally production-oriented, with a very small engineer-
ing staff) drew on the aerospace division's extensive
engineering talent, which included 150 metallurgists
alone. As we see below, a successful working relation-
ship eventually developed between the divisions.
  Among the changes made as a result of the analyses
was the substitution of an aluminum alloy, familiar to
the aerospace engineers, for the nine percent nickel
steel used by Moss. Alcoa was then hired to conduct
additional fracture-mechanics evaluations needed to
obtain U.S. Coast Guard and American Bureau of
Shipping approval for the spherical, aluminum tanks.
These tests also provided "objective" evidence in sup-
port of General Dynamics' applications to MarAd for
subsidy and insurance guarantees.
  On 19 September 1972, three affiliates of Energy
Transportation, Inc., signed a contract with General
Dynamics for three 125,000-cubic-meter LNG tank-
ers. The ships will be chartered to Burmah Oil Tank-
ers, Ltd., for carrying Algerian gas to the U.S. east
coast.
  At the end of contract negotiations, a special 10-
man cryogenics group was formed at Quincy Division.
An aerospace engineer with 15 years of experience in
large missile programs at two companies was selected
to head the group. Given a free hand to transfer the
requisite engineers from the company's other divisions
to Quincy, he brought together a team of aerospace
engineers with expertise in cryogenics, electronics, and
thermal dynamics. He also included several marine
engineers in a conscious effort to integrate the new
team with the shipyard's personnel. Ironically, a major
problem in transferring the aerospace engineers to
Quincy was convincing them to move from San Diego
to the east coast. Of the 10 men asked to transfer, only
two consented.
  The potential transfer of aerospace technology to
the LNG tanker was realized through ties of the cryo-
genics group to the company's aerospace-oriented divi-
sions. The cryogenics group redefined specifications
and supervised procurement of all cryogenic hardware
for the three ships. (Only the piping was provided by
Quincy.) In many instances, the team's aerospace ex-
perience was used. For example, data and methods
developed  in  the space  program  to calculate  heat
  
NUMBER 3

25



transfer coefficients were used in selecting heat ex-
changers. "Redundant" pumps assured fail-safe oper-
ation of the cargo system.
  Similarly, much of the hardware selected for the
tanker reflected advancements in aerospace use be-
yond basic concepts developed in other cryogenics in-
dustries. Examples include submerged cryogenic
pumps and heat exchangers (discussed above), strain-
ers, compressors, capacitance probes, densimeters,
stress gauges, Teflon gaskets, butterfly valves, and
seals. Suppliers were both aerospace and nonaerospace
companies; in a few instances, the former were elimi-
nated from competition on the basis of price; in
others, companies with aerospace orientation were se-
lected over the low bidder in technical review.
  High costs, design complexity, and lack of approval
by the Coast Guard militated against use of some of
the more advanced aerospace developments. Notable
among them were superinsulations, composite mate-
rials (e.g., boron fiber and graphite fiber for strength-
ening and tank supports), and flexible metal hoses
(for lines to spray LNG into cargo tanks). Three of
the superinsulations applicable to LNG tankers were
noted above. General Dynamics had also developed a
superinsulation for satellites under a NASA contract.
This high-performance, multilayer insulation, with
small tufts of dacron needles bonded to one side of a
radiation shield, was considered unsuited for use in an
LNG tanker. Insulation for satellites demands reliabil-
ity over a short life, minimum weight, and reflectivity.
The LNG tanker, on the other hand, requires insula-
tion for convection and conduction of heat, reliability
for up to 20 years of repeated use, and few weight limi-
tations. Initial paper studies and tests to verify the in-
sulation that was finally used in the General Dynamics
tankers (a cheaper, more conventional urethane sub-
stance) were developed by the Convair Division, how-
ever, Some basic engineering principles used in the
company's superinsulation were also applied.
  The aerospace background of the cryogenic team
was further evidenced in its greater tendency toward
formal documentation, quality control, full-range
component and system pretesting, and computerized
testing than was usual in shipyards. In some instances,
these differences also grew out of differences in the
marine and space environments. For example, exten-
sive testing and quality control were often required for
untried aerospace components, because it is impossi-
ble to retrieve and repair failed hardware in flight. By
contrast,   the shipbuilding industry benefitted from

time-proven components and the ability to perform
maintenance afloat. The cryogenics group at Quincy
did pretest the LNG tanker's piping and valve system
(though not by component) to detect potentially seri-
ous leaks. A computer program developed for aircraft
analysis by Convair of Fort Worth was also used to de-
termine the temperature distribution and the quality
of steel in the inner hull.
  During construction the cryogenics group was rela-
tively "self-contained" and had only "selective" con-
tacts with the rest of the shipbuilding staff. A partici-
pant commented that the relationship between the two
at the yard was characterized by the aerospace engi-
neers trying to convince the naval engineers that the
LNG tanker wasn't a great problem, while the latter
assured the cryogenics group that they "didn't have to
be so particular."
  The cryogenics group maintained considerable con-
tact with former colleagues at Convair. "Whenever we
had a problem we just called up Convair, and they
helped solve it. It was a joint effort."23 Among the
services the sister division rendered (in addition to
those already cited) were judgments on acceptability
of welding of the aluminum spheres, and analysis and
design studies for the tank production line. (Convair's
experience in designing tools to assemble large com-
ponents for the space shuttle and DC-10 fuselages was
cited as extremely relevant to these services.) The last
task grew out of problems with Pittsburgh Des Moines
Steel (PDM), to whom General Dynamics had .sub-
contracted the fabrication of the 120-foot-diameter
spherical tanks. Before the first set of tanks was fabri-
cated, PDM defaulted on its contract. In December
1974, General Dynamics purchased the site of PDM
tank construction and took over production of the
spheres.
Since 1972, General Dynamics has received orders
from Energy Transportation Corporation for five ad-
ditional 125,000 cubic meter LNG tankers. Because
these ships will enter the Indonesia-Japan route, they
are not as vulnerable to U.S. regulatory controls.
Nevertheless, all are being built to the same fun-
damental design specifications as the first three. No
attempt has or is likely to be made to introduce more
advanced (aerospace) technology. According to a
Quincy Shipyard executive, this heeds the shipowner's
desire for standardization and the shipbuilder's reluc-
tance to try technology that has not "proven itself to-
tally reliable in the context of commercial usage." 24
In the meantime, however, the project for which the

26

SMITHSONIAN STUDIES IN AIR AND SPACE



first three General Dynamics tankers were ordered
faces an uncertain fate. The importing utility began
filing applications for approval early in 1973. But the
FPC failed to rule before the contract deadline
(1975), and the Algerian government-owned supplier
"cancelled" the contract. Renegotiation, which is ex-
pected to result in an increase in the price of the gas of
up to 300 percent, and reapplication for approval are
now underway.
"Inhibitors" and "Promoters" in the Process of
Technology Transfer
  The LNG tanker case study amply illustrates some
of the major influences in the transfer of technology to
secondary uses. It encompasses major "inhibitors" and
primary "promoters," which are common to other
cases as well.
  By far the most important constraints to transfer-
ring advanced aerospace technology to LNG tankers
are the considerable financial and safety risks. The
huge capital investments required for tankers and
shoreside complements to a natural gas import project
have far-reaching implications for technological inno-
vation in the industry. Shipowner, shipbuilder, and
ship lessee alike are understandably reluctant to jeop-
ardize major outlays and potential profits by intro-
ducing technology that may be unpredictable or cost-
ineffective in its new environment.
  Sophisticated aerospace technology is frequently
more expensive than conventional technology. Aero-
space technology might afford some improvements
over items currently in use, but if it does not provide
improved reliability or other benefits to outweigh the
increase in price, transfer is inhibited. Internal super-
insulations, new metal alloys, and fibers were cases in
point. The leader of General Dynamics' cryogenics
group put it succinctly: "If it were economical, we'd
gladly use it." 25
  Another critical constraint to transferring technol-
ogy is the institutional net that engulfs any LNG
tanker destined for U.S. trade. The list of federal,
state, and local regulatory agencies involved is long; it
has been only partially detailed above. The safety issue
is of direct concern to most of those involved in ap-
proving an LNG project.26 Most participants take the
conservative approach. For example, the U.S. Coast
Guard, which sets standards and certifies the safety of
any U.S. flagship, published standards for LNG tank-
ers in 1965 based on the contemporary state of the art.
Updated after 10 years, the standards still do not re-

flect many of the technological advances made by the
aerospace sector in the interim.27
  For its part, the PFC not only regulates economic
aspects of gas import projects but, in approving termi-
nal facilities, must also meet the requirements of the
Environmental Protection Act. For the first major
U.S. LNG import project, the FPC took 49 months to
render the necessary approvals. In the past few years,
procedures attendant upon filings, hearings, amend-
ments, and approval of environmental impact state-
ments have increased, and they now constitute one of
the major hurdles in the entire process. This is at least
partly attributable to the 1973 explosion of a Staten Is-
land LNG storage tank, which killed about 40
people.28 The result of the accident has been the pro-
liferation of organizations involved; intensification of
their conservatism; and reintroduction of a plethora
of social, environmental, and political considerations.
  Legal arguments at FPC hearings now focus primar-
ily on selecting sites for terminal facilities to which for-
eign LNG can be delivered. Emphasis is often placed
on identifying "excessive risks" involved in such occur-
rences as a sinking or major collision involving an
LNG tanker in a busy harbor. (Pollution does not fig-
ure prominently in the issues.) Even the Coast Guard's
assurances of the safety of the ship, a letter-of-
compliance program for all ships entering U.S. har-
bors, contingency plans for special navigational aids,
and the relative safety of LNG tankers as compared to
oil tankers in the respects of fire, pollution, and
grounding have not blunted opposition to the
terminals—neither have appeals on the basis of na-
tional energy needs.29
  Illustrative of this maze of bureaucratic and social
barriers are the two-year old hearings for approval of
several LNG tanks in the Staten Island area for a large
import project to New York harbor. Through
mid-1975 the hearings attracted over 80 legal partici-
pants and interveners and generated 50 to 60 volumes
of pleadings. There was no end in sight and no clear
answer even to the status of the procedures.30 In addi-
tion, fallout from the case was felt in almost every
other LNG project where approvals were pending.
Those projects where approvals had already been
granted were being reopened.31 Of the latter, one was
reopened because of an increase in price imposed by
the Algerians.
  The outlook for importing LNG and ordering LNG
tankers is unquestionably affected by these regulatory
hassles. One executive commented after two years of
awaiting FPC approvals for a large import project:
  
NUMBER 3

27



"We do not see how we can hold the project together
much longer." 32 Another observed that delays and
uncertainties in regulatory procedures present "im-
measurable risks" that have already left behind a
"series of abortive projects." 33 In spite of this gloomy
picture, however, some of those fighting the battle
against LNG terminals admit that the big economic
interests at stake and the country's energy require-
ments will prevail; LNG import projects will probably
go forward eventually; and the most they can hope for
is more stringent safety regulations, a strong national
policy on siting,34 and substitution of other fuels for
natural gas.
  The economics of importing LNG and the inconsis-
tencies in the national energy policy (or lack thereof)
are also major hurdles for tankers designed to carry
only that fuel. As noted above, FPC controls on natu-
ral gas have kept both the price and the supply of do-
mestic natural gas down. At the wellhead, natural gas
was priced at the end of 1975 at about 50 cents per
thousand cubic feet and was delivered at about $2.25-
$2.75, depending on the distance from the source of
supply. Algerian gas had risen to about $1.20-$1.40
per thousand cubic feet, which handling, shipping,
conversion, reconversion, and terminal costs brought
to between $3.00-$3.50 (in the best case of a fully
operational project). Import LNG was therefore mar-
ginally competitive with domestic gas.35 The econom-
ics of importing LNG, however, are subject to interna-
tional politics, unpredictable operating costs, relative
prices of competitive fuels, and —very significantly—
the potential for increasing use of other energy
sources. The widening gap between domestic demand
and supply of natural gas is a strong incentive for U.S.
energy interests to pursue efforts to import LNG. But
an adverse change in any of the factors identified
above could seriously alter the outlook for large-scale
LNG import projects.
  Fundamental differences between the aerospace
and marine environments also impede transfer of
technology between the two. The shipbuilding indus-
try has long been conservative. In its highly competi-
tive world marketplace, cost is a big factor, profit
margins are small, and skilled labor is expensive. Basic
technology, with some exceptions, has been largely the
same for decades, providing assurance of rugged,
long-lived equipment.
  By contrast, the aerospace environment is fast mov-
ing and progressive. Cost has historically been a sec-
ondary consideration. (While this is changing in the
competitive arena of commercial aircraft, it was char-

acteristic of the manned space program and of many
of the military rocket or missile and aircraft pro-
grams.) The aerospace environment is often unforgiv-
ing; it demands light weight, miniaturization, and
high reliability for a relatively short life. These special
demands have been met with research and develop-
ment programs and extensive quality control and pre-
testing (all of which are very costly). At the edge of
new developments and advanced technology, the aero-
space industry has been prone to improve and inno-
vate rapidly (alledgedly sometimes to the extent of in-
corporating technical advances for their own sake).
The effect of these fundamental differences has been
to make transfer of aerospace technology to the mar-
ine environment difficult.
  In spite of the above constraints, there were several
factors that served to stimulate transfer of aerospace
technology to the LNG tanker. Most significantly, by
far, was the use of engineers who effected transfer.
The indispensable role of these promoters was clearly
evident in the LNG tanker case. Whether forced or
merely nudged by corporate management, when the
aerospace engineers were transferred to the shipbuild-
ing division, they took their technological bags with
them. They drew on their aerospace experience and
methods, and they maintained close contact with for-
mer aerospace colleagues, who provided advice and
assistance. The successful transfer of technology must
be attributed in the first instance to the transfer of
technologists. The essence of most of the technology
transferred—analytical methods and engineering
approaches—resided in the individuals, not in the
devices.
  Other major influences that promoted the transfer
of aerospace technology to LNG tankers were the
growing national need for energy; the shared interest
in cryogenic technology between the aerospace and the
LNG tanker industries; federal government interest in
stimulating development of U.S. LNG tankers and,
coincidentally, in utilizing cryogenic technology de-
veloped for the space program; and the relationship
between Quincy Shipyard and a large aerospace con-
glomerate.
Conclusion
  The successes in transferring aerospace technology
to the LNG tanker are testimony to the decisive role
that the technologist plays. They also reflect the im-
portance to application in other environments of many
aerospace advances in technology. The need for addi-
tional  energy  resources,   the  influence  of financial
  
28

SMITHSONIAN STUDIES I



interests, and the momentum of tanker production
programs may prove effective promoters of future
transfers to LNG shipbuilding.
  At the same time, a poor prognosis for future trans-
fers may prevail: The price of foreign natural gas has
soared; national priorities have shifted from clean
energy to self-sufficency; MarAd has virtually elimi-
nated its entire cryogenic support program; indicators
in the tanker industry are bearing out predictions of a
decline in demand for LNG tankers by 1990; the
widespread side effects of the 1973 Staten Island ex-
plosion have discouraged potential gas importers; and
the FPC has not yet (1977) given final approval to a
single import project for which any of the 16 LNG
tankers under construction in the United States are
designated.
  
Whatever the future of the LNG tanker, the experi-
ence gained in this case provides useful lessons for
understanding and driving the technology transfer
process. Successful transfer, for example, clearly de-
pends to a large degree on the existence and support of
several effective promoters: innovative and dynamic
people; strong financial support; identifiable needs;
relevant technological solutions; cooperation between
key institutions and individuals; tolerance of risk; and
demanding specifications of performance.
  There are several significant factors that consistent-
ly inhibit transfer: cost, obstructive policies, incom-
patible environments, political influences, and inap-
propriate technology are a few. Understanding these
factors may well lead to the critical guidelines neces-
sary for achieving successful transfer of technology.
  
Notes

  1. Department of Commerce, Maritime Administration, LNG
Tank Designs (Washington, D.C, December 1972), NMRC-272-
230000-R1.
2. Ibid.
  3. Ibid.; William du Barry and Alfred H. Schwendtner, "LNG
Carriers: The Current State of the Art," paper presented at annual
meeting in New York of the Society of Naval Architects and Marine
Engineers, 11 and 12 November 1971; Report: LNG Conference,
29-30 March 1972 (Washington, D.C.) sponsored by U.S. Mari-
time Administration and Shipbuilders Council of America; Analysis
of LNG Marine Transportation, report prepared by Booz Allen
Applied Research for the Maritime Administration, November
1973, volumes I, II.
  4. Department of Commerice, LNG Tank Designs; du Barry and
Schwendtner.
  5. George D. Caramero, Jr., El Paso Natural Gas, "Energy, Mon-
ey and Ships," remarks, Seatrade Conference, New York, 26-27
March 1973.
6. Ibid.
  7. Interview, Philip L. Ross, assistant to the general manager,
Market Development, General Dynamics, Quincy Shipbuilding
Division, 18 October 1974, 11 and 20 August 1975, and letter with
enclosed comments, 29 April 1975.
  8. "Benefits Briefing Notebook" prepared by the Denver Re-
search Institute for NASA Technology Utilization Office, Septem-
ber 1974 (review draft) ; Denver Research Institute, Cryogenics, A
Technology Transfer Profile, Applications of Aerospace Technol-
ogy in Industry (December 1971), prepared for NASA; Denver Re-
search Institute, A Case Study in Technology Utilization: Fracture
Mechanics (May 1972), prepared for Technology Utilization Office
of NASA; interview, Rolf D. Glasfeld, chief development engineer,
General Dynamics, Quincy Shipbuilding Division, 18 October 1974,
and review comments, 29 April 1975; interview, Ross Buck, cryo-
genics engineer, General Dynamics, Quincy Shipbuilding Division
(formerly with Convair Division), 18 October 1974; telephone
interview, Alan Schuler, manager Cryogenics Engineering, Quincy
Shipbuilding Division, General Dynamics (formerly with Convair
Division), 11 and 18 August 1975; Martin Marietta, "Capillary In-

ternal Insulation System" (technical data report, March 1973);
Rockwell International, Space Division, Wet Wall Insulation System
for LNG Tankers (June 1973) ; interview, Richard Breed, Materials
Division, National Bureau of Standards, Cryogenics Division,
Boulder, Colorado, 19 September 1974.
  9.	Analysis of LNG Marine Transportation; Martin Marietta,
"Capillary System"; Rockwell, Wet Wall Insulation; Cryogenic Re-
search at Marshall Space Flight Center, Research Achievements Re-
view, volume IV, report number 2 (May 1970-May 1972), NASA
Technical Memorandum TM X-64561; memorandum, University
of Denver, Research Institute (to NASA) TEF 361, case number
10350221, 3 September 1974, subject: recontact interview with John
Barnes, Rockwell International Corporation.
  10. Department of Commerce, LNG Tank Designs; du Barry
and Schwendtner; Analysis of LNG Marine Transportation.
11. Schuler interview.
  12. L. R. Smith (project engineer, J. S. Carter), "Submerged
Pumps for LNG Sendout," paper presented to the American Gas
Association, Distribution Conference, May 1968, Houston, Texas.
13. Schuler interview.
  14. None of the other LNG shipbuilders, i.e., Avondale, New-
port News, and Sun Shipbuilding and Dry Dock, had direct aero-
space connections.
  15. Unless otherwise noted, all information directly related to the
General Dynamics (Quincy Division) LNG tanker program was pro-
vided through the previously cited personal interviews and telephone
conversations with Ross, Glasfeld, Buck, Schuler, and through in-
terview, Edward Heineman, vice president of Engineering and Pro-
gram Development, General Dynamics (retired), 23 October 1974.
  16. Nine U.S. LNG tankers had been subsidized through 1975,
the last three at the rate of 16.5 percent.
  17. John J. Nachtsheim, deputy assistant administrator for R&D,
Maritime Administration, "The Maritime Administration LNG
R&D Program,' paper presented to LNG Conference, 29-30 March
1972, Washington, D.C.
  18. Andrew E. Gibson and Marvin Pitkin, Maritime Administra-
tion, "Developments in LNG Transportation in the U.S.," paper de-
livered to the Third International Conference and Exhibition of Li-
  10. 
NUMBER 3

29



quid Natural Gas, 26-28 September 1972, Washington, D.C.
  19. Breed interview; interview, Marvin Pitkin, assistant admin-
istrator for commercial development, Maritime Administration, De-
partment of Commerce, 13 September 1974; interview, Joseph Seel-
inger, LNG program manager, Maritime Administration, 19
September 1974; Cryogenics Division, National Bureau of Stand-
ards, Liquefied Natural Gas Research at the National Bureau of
Standards, progress report for the period 1 January to 30 June 1974,
D. B. Mann, editor, Washington: GPO, 1974, NBSIR 74-373.
  20. Not entirely for the LNG tankers but partly to reduce the la-
bor component in the shipyard generally.
21. Glasfeld interview.
  22. The difficulty and need to "force" interdivisional transfer was
also noted in the cases of Boeing, Westinghouse, and Philco-Ford.
23. Schuler interview.
24. Ross interview.
25. Schuler interview.
  26. The process of gaining approval for an LNG import project is
focused primarily on shoreside facilities, but because the tanker's
ability to operate is clearly a function of their approval, the entire
process and all its participants have as great an impact on any tank-
er in U.S. trade as Coast Guard certification of the tanker design
itself.


  27. "LNG Safety: An Overview," draft of paper presented by
Julius Feldman to Ocean 75 Conference, 22 September 1975, and
interview with Julius Feldman, Office of Research and Develop-
ment, USCG, 23 September 1974.
  28. The explosion occurred during maintenance of an empty
tank (which had been used to store LNG briefly in 1971-1972) and
was attributed to faulty construction of the insulation.
  29. "LNG Safety: An Overview"; telephone interview. Bruce
Terris, attorney for Congressman John Murphy (D-NY) in Staten
Island LNG Terminal Hearings, 3 September 1975.
30. Terris interview.
  31. Federal Power Commission, news release, number 21654, 20
August 1975, on LNG applications (Washington, D.C.).
32. Caramero, "Energy, Money and Ships."
33. Ross interview.
  34. A bill pending before Congress required the Commandant of
the Coast Guard to select "appropriate" sites for LNG storage ter-
minals (as opposed to current selection of such sites by the com-
panies involved), HR 4440, bill introduced to House of Representa-
tives, 6 March 1975, by Congressman John M. Murphy (D-NY).
  35. Telephone interview, William Jaques, vice president Sales
and Rates, Algerian Gas Company, 2 October 1975.
  27. 
The Delphi Method of Technological Forecasting
J. Gordon Milliken

EDITOR'S COMMENT
  Our third study describes a social scientific invention. The meth-
od called "Delphi" was created to predict developments in weaponry
after World War II, and its close alliance to this technology has
motivated us to include it here. Nevertheless, Delphi is removed
from the aerospace industry one step further than the technology
discussed in the previous causes. Indeed, while the method was devel-
oped at the Rand Corporation— a military and aerospace industrial
"think tank"—it has been used increasingly in industries that have
no interest in aerospace hardware or methods. Delphi is a means of
prediction by surveying opinion of a panel of experts, to create a col-
lective judgment. Its defining characteristic is a mechanism of
"feedback" by which diverse expert opinions are induced to con-
verge toward a consensus.
  J. Gordon Milliken, the author of this study of Delphi, is a man-
agement scientist at Denver Research Institute and headed a team of
researchers that produced nine studies for the National Air and
Space Museum. This study and the two that follow were extracted
from their unpublished report of 1975, Aerospace Contributions to
Management. He has exhaustively searched the literature of Delphi,
which is immense, and he has conducted interviews with founders
and users of the method. The major portion of his study is devoted
to a list of many uses to which Delphi has been put, but his most in-
teresting research is embodied in two sections: "A Study of How
Delphi is Used" and "The Validity of Delphi."
  In the first of these, Milliken has chosen * Delphic study that
preceded a conference on education of management at the Universi-
ty of Colorado. He has described the method of questioning, the
iteration and feedback, and the judgment of the participants that
their use of Delphi was successful. He notes that the discussion of the
method in the conference was itself a mechanism of diffusion of
knowledge of Delphi.
  In the second of these, Milliken's research suggests that the history
of Delphi is like the history of some new methods in the exact
sciences. A major synthetic accomplishment— the theory of ordinary
differential equations, for example—is followed by widespread use
and modification of the new method. Soon, however, a critical reac-
tion sets in, for despite its success, the method is built on uncertain
foundations. The search for proofs of existence, uniqueness, and
reproducibility constitutes a phase of growing rigor, and in Delphi it
takes us to the present. We find in Milliken's documentation that
many have judged the method to have been widely misused. But the
range of applications expands, and the advocates of Delphi predict
more frequent use.
J.  Gordon Milliken, Denver Research Institute, Denver, Colorado
80210.
  
Planners in education and government will no doubt use Delphi
to reinforce their preliminary judgments, but corporations may have
less use for it. There are at least two reasons for this. First, a chief ex-
ecutive or board of directors may decide the course of the firm, ex-
pecting that it will change also the environment in which it conducts
business. Delphi is unlikely to anticipate such a change. A company
that controls a large portion of the supply to the market in which it
competes changes the behavior of the entire market when it changes
its level of production. The monopolistic company can quickly
render the perception of a Delphi panel invalid, because the panel,
even if it sits within the company, is not privy in advance to major
policy decisions. Delphi smooths out discontinuities, discounts radi-
cal changes, and thus is insensitive to the important effects of cor-
porate policy on economic and technological development. Second,
Delphi imposes an objective scale on highly subjective, often quali-
fied statements. Yet a corporate executive who solicits expert opin-
ion wants to weigh the ramifications of the issue he or she is about to
decide. Delphi gives him less information than full expert opinions,
and it usurps his responsibility to make a decision that may deter-
mine great profit or loss.
  Even in industry, however, routine decisions in planning can
come from Delphi. With the essential addition of a feedback mech-
anism, Delphi is a significant alternative to simple polling. At the
cost of time and of money it provides a new kind of information, a
consensus of sorts, which may be carefully used to advantage.
  Delphi is a method for acquiring collective predic-
tions of events from many informed judges.1 When
lack of objective data rules out conventional techno-
logical forecasting by extrapolation, for example,
Delphi is used to develop a consensus of experts.
Where an adequate consensus cannot be reached,
Delphi can at least reveal polarization of opinions be-
tween two factions. From its early use at Rand and in
other research groups working for the U.S. Air Force,
Delphi has spread to government agencies, corpora-
tions, and academic institutions. The Delphi method
of technological forecasting is one of very few tech-
niques for planning which is new, not the product of
adaptation or improvement of a prior method.
  Delphi entails the repeated interrogation of re-
spondents (usually chosen for their expertise) regard-
ing their expectations or understandings. They are
queried separately, to avoid biases of authority. The
  
30

NUMBER 3

31



DENSITY
.8
median or distribution of responses is then furnished
to the respondents, who may stand by their judgment
or alter it. In some cases, especially for deviant opin-
ions, statements of justification are sought, and these
also are furnished to the respondents to further stimu-
late original judgments.2 Finally, the estimates are
stated in the form of a consensus judgment.
  Norman Dalkey, a coinventor with Olaf Helmer of
the Delphi technique, describes the two basic assump-
tions that underlie Delphi inquiries:
  (a) In situations of uncertainty (incomplete information or in-
adequate theories) expert judgment can be used as a surrogate for
direct knowledge. I sometimes call this the "one head is better than
none" rule.
  (b) In a wide variety of situations of uncertainty, a group judg-
ment (amalgamating the judgments of a group of experts) is pref-
erable to the judgment of a typical member of the group, the "n
heads are better than one" rule.'
  The use of Delphi avoids some common problems of
decisionmaking in a committee or special task force.
The technique eliminates judgments based on "he who
shouts the loudest," and it weakens inhibitions some-
times held by group members of lesser rank or reputa-
tion.4 Further, it allows members of the group to
abandon earlier judgments that are later seen to be er-
roneous, without the embarassment of reversing pub-
licly held opinions. This is indicated by the tendency
of opinion to converge after several rounds of anony-
mous response and feedback. Delphi also tends to
avoid the problem, encountered in group discussions,
of confusing the main issue with irrelevant argument
among members of the group. It also controls group
pressure, which strongly influences members of a
group to compromise their sincere judgments.
  Norman Dalkey points out that Delphi procedures
have been designed to minimize the undesirable effects
of group interaction described above:
The [Delphi] procedure has three distinctive characteristics:
Anonymity
Controlled feedback
Statistical "group response"
  Anonymity is a device to reduce the effect of the socially dominant
individual. It is maintained by eliciting separate and private answers
to prepared questions. Ordinarily, the procedure is carried out by
written questionnaire; on-line computers have been used for some
exercises. All other interaction between respondents is through for-
mal communication channels controlled by experimenters.
  Controlled feedback is a device to reduce noise (among other
things). A Delphi exercise will usually consist of several iterations
where the results of the previous iteration are "fed back" to the
respondents, normally in summarized form.
  As a representative of the group opinion, some form of statistical
index is reported. For cases where the group task is to estimate a
numerical quantity, the median of individual estimates has turned

out to be the most useful index tried to date. Thus, there is no par-
ticular attempt to arrive at unanimity among the respondents, and a
spread of opinions on the final round is the normal outcome. This is
a further device to reduce group pressure toward conformity.5
  Some understanding of the effect of iteration can be
obtained from a comparison of the graphs in Figure 1,
taken from Dalkey's The Delphi Method: An Experi-
mental Study of Group Opinion.6 Figure 1 (top)
shows that the distribution of first-round answers to
numerical questions that have specific answers ap-
proximates a log normal curve. Figure 1 (bottom),
taken after one round of feedback and a new response,
shows that the answers have shifted dramatically
toward the mean, indicating a convergence of answers
toward the (average) response of the group.
Dalkey summarizes the Delphi methodology:
  A typical exercise is initiated by a questionnaire which requests
estimates of a set of numerical quantities, e.g., dates at which tech-
nological possibilities will be realized, or probabilities of realization
by given dates, levels of performance, and the like. The results of the
first round will be summarized,  e.g.,  as the median and inter-
n 06SECVE0
— LOG  NOCMAL
2.5
.5
3.0
1.0	1.5	2.0
NOGMALIZED   ESTIMATE
DENSITY
•8r

.2-i
.5
2.5
3.0
1.0	1.5	2.0
NORMALIZED   ESTIMATE
FIGURE 1.—Distribution of initial answers  (top);  distribution of
second-round answers (bottom).
1      -(LOGx)'

32

SMITHSONIAN STUDIES IN AIR AND SPACE



quartile range of the responses, and fed back with a request to revise
the first estimates where appropriate. On succeeding rounds, those
individuals whose answers deviate markedly from the median (e.g.,
outside the interquartile range) are requested to justify their esti-
mates. These justifications are summarized, fed back, and counter-
arguments elicited. The counter-arguments are in turn fed back and
additional reappraisals collected. This basic pattern has, of course,
many possible variants, only a few of which have been tried.'
  Appendix 1 is the set of typical instructions for a
Delphi experiment using questions for which the
answers are known to the experimenters.8 Experiments
using such almanac-type questions as "What was the
number of telephones in Africa in 1962?" or "How
many married students were enrolled in high school in
the US in 1965?" can test the accuracy and con-
vergence of Delphi. Where confirmation is possible,
experimenters have found that opinions tend to con-
verge during the experiment, and generally the me-
dian response moves in the direction of the true
answer. Where confirmation is not possible, one can
say only that opinions do converge during the
exercise.9
  Appendix 2 contains examples of the almanac-style
questions used by Norman Dalkey and his Rand col-
leagues in Delphi experiments with students of the
University of California at Los Angeles. The size of
group error in the responses is also shown.
  One additional feature of present Delphi procedures
should be mentioned. Respondents are usually re-
quested to rate their self-confidence for each response.
Several different kinds of self-ratings have been tried:
ranking the questions in the order of the respondent's
judgment of his competence to answer them and esti-
mating self-confidence with respect to some reference
group. In general, there has been no significant cor-
relation discovered between self-ratings and "indi-
vidual" performance for estimates that can be con-
firmed. It has usually been possible, however, to select
a subgroup of relatively more confident individuals
whose performance has been slightly, but consistently,
better than the group as a whole. In one very thorough
study, the improvement was obtained only by combin-
ing two self-rating indices—the ranking of questions
and absolute estimates of confidence.10
  Much work with Delphi has been experimental, but
the results of the experiments show considerable prom-
ise for its regular use. Dalkey has found in experiments
with college groups that the group opinion improves
after Delphi iteration, whereas the consensus becomes
less accurate if group discussion is allowed.11

History of Delphi
  Technological prophecy has existed throughout his-
tory.12 Technological forecasting, a less intuitive and
more disciplined effort, was given major impetus dur-
ing World War II as the U.S. Army Air Force began
planning for the future. In 1944, Theodore von Kar-
man and others published Toward New Horizons,
which related expected technological growth to the
future Air Force.
  At the end of World War II, many persons in gov-
ernment were convinced that it was important to con-
tinue the collaboration between scientists and the
military established during the war. Civilian advisors
to the Secretary of War proposed a new kind of orga-
nization to serve the national security and the public
interest. The Air Force let a research contract to
Douglas Aircraft Company of Santa Monica for Proj-
ect "RAND." 13 In 1948, it was decided to separate
RAND from Douglas. With Ford Foundation funding,
and a distinguished nongovernmental board of trust-
ees, the Rand Corporation was formed "to further and
promote scientific, educational, and charitable pur-
poses, all for the public welfare and security of the
United States of America." Its first and principal con-
tract, for the conduct of Air Force Project RAND,
called for "a program of study and research on the
broad subject of intercontinental warfare, other than
surface, with the object of recommending to the Army
Air Forces preferred techniques and instrumentalities
for this purpose." 14 The first RAND project, com-
pleted and reported in May 1946, was a study of the
feasibility and utility of an earth-circling space ship.
  In the early years of Rand, experiments were con-
ducted on the techniques of technological forecasting.
Norman Dalkey mentions an intermittent series of
studies on using group information effectively: "The
early studies were concerned mainly with improving
the statistical treatment of individual opinions." 15 A
key event in the development of forecasting techniques
occurred in 1950 when a group of statisticians did a
study to determine the results of pooling horse-race
handicapping opinions. Though one would lose after
betting on a number of horse races, one would not lose
as much when following the opinions of the pooled
handicappers as when one followed the advice of one
handicapper. Because of the "negative" results (that
the better would lose), the significance of this study
was not recognized immediately, and it was never pub-
lished. Another extensive study of short-range predic-
tions16 had indicated that some properties of individ-
  
NUMBER 3

33



ual estimates (precision, definiteness) could be used to
correct the accuracy of short-term predictions and
that background information (as measured by a
standard achievement test) had a small but significant
influence on the accuracy of predictions. Both of these
effects, however, were fairly washed out by combining
estimates into a group prediction.
  The importance of these results became apparent as
Rand scientists tried to determine the probability of
general war in an evaluation ol future enemy weapons
for the Air Force. Early in their research it became
clear that technological developments should be pre-
dicted, and to this end they decided to experiment
with intuitive judgments of physical and social scien-
tists.
  Faced with a U.S. Air Force request for a classified
study of bombing requirements, and needing techno-
logical forecasts, Norman Dalkey and Olaf Helmer in
1951 built upon this early work and conducted the first
true Delphi study.17 Their experiment was from the
viewpoint of a Soviet strategic planner— to apply ex-
pert opinion to select an optimal group of U.S. targets
and to estimate the number of atomic bombs required
to reduce the munitions production by a required
amount." Dalkey and Helmer's important innovation
was the introduction of iteration with controlled feed-
back. The set of procedures that evolved from this
work has received the name "Delphi" —a misleading
appellation in Dalkey's opinion, since his colleagues
and he believe that there is little that is oracular about
the methods.19
  Dalkey commented that all of the work done on
Delphi at Rand from 1952 to 1962 was purely aero-
space activity; some of the studies are included in
Rand's 1974 selected bibliography.20
  The Delphi procedures received a great boost in
general interest in 1964 with the publication of Gor-
don and Helmer's study of forecasting technological
events.21 This study produced predictions as much as
50 years into the future, dealing with six areas: scien-
tific breakthroughs, population growth, automation,
space progress, probability and prevention of war, and
future weapons systems. The study happened to coin-
cide with a surge of interest in long-range forecasting,
with an attendant interest in the systematic use of ex-
pert opinion.22
  Throughout the 1960s, Delphi was used predomi-
nantly in aerospace industry and the military. The
U.S. Air Force funded Rand's continuing research and
conducted some experiments of its. own. The Air Force

Systems Command23 and Office of Scientific
Research24 conducted pilot studies of social and tech-
nological forecasts. The technique was employed in at
least two doctoral dissertations in the mid-1960s,25 and
its use in military and commercial applications be-
came more extensive during the late 1960s.26
  There were several important nonaerospace appli-
cations, however. In the late 1960s, the Institute for
the Future (IFF) conducted a series of studies for the
state of Connecticut, which were designed to examine
the state's future and to aid long-range planning for
Connecticut.27 Related work was conducted by IFF for
the Electric Power Research Institute.
  In 1969 Dalkey stated in a Rand memorandum, "At
present, it is difficult to obtain a clear picture of how
widespread the applications are; but a crude guess
would put the number of studies recently completed,
underway, or in the planning stages at well over a
hundred." 28
Applications
  In a review of Delphi, Olaf Helmer describes the
proliferation of uses of the technique:
  While its principal area of application has remained that of
technological forecasting, it has been used in many other contexts in
which judgmental information is indispensable. These include nor-
mative forecasts; the ascertainment of values and preferences; esti-
mates concerning the quality of life; simulated and real decision-
making; and what may be called "inventive planning," by which is
meant the identification (including invention) of potential meas-
ures that might be taken to deal with a given problem situation and
the assessment of such proposed measures with regard to their feasi-
bility, desirability, and effectiveness.*9
  There are two basic areas of application of Delphi,
according to Helmer,
one of which has to do with inputs into research, the other with the
utilization of research output: The first such application consists in
the employment of Delphi surveys to provide judgmental input data
for use in studies in the social-science area, in cases where hard data
are unavailable or too costly to obtain. For instance, in a study of
socioeconomic strategies for a developing country, reliance on
judgmental information supplied by area specialists may be an ade-
quate substitute for hard, but nonexistent, data. The other major
application of Delphi, of which numerous examples already exist, is
to the process of gathering of expert opinions among the nationwide
"advice community" on which governmental decision-makers fre-
quently rely. In this mode of application, Delphi can be of consid-
erable utility, both by systematizing the process and by lending
greater objectivity to its "adversary" aspects.50
  TRW, Inc., a consolidation of the aerospace firm of
Ramo Woolridge with the conventional manufacturer
  
34

SMITHSONIAN STUDIES IN AIR AND SPACE



Thompson Products, Inc., has been an early user of
technological forecasting "that can provide manage-
ment with the distance vision it now requires to do its
job . . to help management improve its decision-
making processes by providing immediate information
on probable and significant future trends." S1 The
primary applications of technological forecasting at
TRW were to plan strategies for research and develop-
ment and to identify opportunities to develop new
products. Harper North and Donald Pyke of TRW
commented:
  We decided to employ a modification of the Delphi method . .
developed and used by the Rand Corporation. This method seeks to
take full advantage of the "committee" approach to forecasting,
while avoiding some of the disadvantages of the typical brain-
storming session ....
  For the most part, our initial attempt to generate a long-range
forecast of the technological future was a test of the feasibility of us-
ing the Delphi method in an industrial environment. It consisted of
an experimental survey, finally completed in June 1966, which led to
a 50-page proprietary document titled, "A probe of TRW's future:
the next 20 years.*' This document listed some 401 technical events,
sorted into 15 categories, which a single panel of 27 experts spe-
cializing in a broad spectrum of technologies felt would occur by
1986 and which they believed would have significant impact on the
company's products, services, and processes. This experiment later
became known as Probe I. Before launching a second stage, Probe
II, the 15 categories of events were refined to focus more closely on
our current interests."
  The TRW experience with Delphi proved not to be
as successful as North and Pyke expected in 1969.
Wayne Boucher, a former Rand colleague of Helmer
and Dalkey's and now with The Futures Group, stated
in an interview that TRW's Probe work was not put to
use as a basis for making decisions on TRW's future
programs, "despite the support of Ramo and others
who committed themselves to making Probe a perma-
nent part of their future. The problem was data over-
load. Yet Delphi did have a great impact on shaping
the conceptual future of technology— and Probe has
apparently served from time to time as a guide to such
thinking."
Norman Dalkey stated in an interview:
  In the area of technological forecasting, TRW Systems used Del-
phi for informational purposes only. They began with a list of 1,800
potential technological events and asked the panel to predict when
they would happen, their feasibility for development, and their sig-
nificance to TRW. They classified events into 20 areas and then
predicted the probability of each event existing within one of several
time periods. TRW planners discovered that they needed to develop
certain intermediate technologies to get to the final point. They
found that intermediate technologies like laser optics were more in-
teresting than the final technologies, and they are now putting de-
velopment money into certain of the intermediate technologies."
  
Besides TRW, Dalkey noted, the two aerospace
firms of Boeing and Northrop have used Delphi for
technological forecasting.
  Boucher noted that The Futures Group has used
Delphi in numerous fields: to forecast the future of the
photographic industry, biological processes, the Japa-
nese eonomy, beverages, food technology, various
energy sources, psychotropic (mind-altering) drugs,
communications equipment, and other subjects. In
the study of communications equipment, The Futures
Group first used a variation of Delphi—interview
Delphi34—to design physical equipment to test its ac-
ceptability by users. Managers of telephone exchanges
were asked to forecast design features desired on a
switchboard. The interview Delphi results were used as
a basis for a multimillion dollar decision. Because of
the considerable financial impact of the decision, it
was made only after the client had spent several
months testing, and failing to refute, the findings of
The Futures Group. Interview Delphi has since been
used by The Futures Group in many other studies. Ac-
cording to Boucher, its advantage over the question-
naire technique is so clear that it has become the pre-
ferred approach to Delphi at The Futures Group.
Other organizations also are now using it.
  Delphi has been adapted to such business-related
matters as the impact of increased leisure time on in-
dustrial product lines.35 Norman Dalkey estimated in
a 1974 interview that during the past five years about
1500 Delphi studies have been undertaken. Most of the
industrial studies, however, have not led to decisions to
act. They seem to be of interest to top management,
but often administrators are unsure of what to do with
the results or are unwilling to commit resources to im-
plement them. Dalkey did not attribute this lack of
impact on industrial policy to any basic flaw in the
Delphi method or to a lack of managerial confidence
in it. Instead, he felt that there was a considerable
distance between the subject of the Delphi exercise
and the subject matter that would be used for a policy
decision. Dalkey hopes that external studies will soon
give way to panels within companies.
  In a recent use of the technique, which is indicative
of Delphi results substituting for data, the California
Air Resources Board surveyed a panel of doctors to
estimate the doses of pollutants that would have cer-
tain effects on incremental percentages of the popula-
tion. This, Dalkey explained, is the most straightfor-
ward application of Delphi, because it is specific.
  On the other hand, government and industry may
use Delphi to identify the goals of their organizations.
  
NUMBER 3

35



Many of these are done in-house as part of long-range
planning, and they are general. In between the two ex-
tremes lies the use of Delphi for technological fore-
casting. According to Dalkey, this most subjective use
of Delphi differs from the more objective applications
in that it is rare that one initially asks the right ques-
tions. A few iterations are necessary solely to define the
questions. Dalkey gave us the example of one organi-
zation that questioned 30 executives at the level of
division chief or above to define the five most impor-
tant goals of the organization. The first questionnaire
brought 127 different responses. Cluster analysis re-
duced the variety of responses to a workable number.
This example demonstrates two problems inherent in
applying Delphi to subjective topics. First, ques-
tionnaires must deal with limited areas in order to il-
luminate relations among many variables. The second
problem lies in determining what is worth forecasting.
In the more subjective areas, those who make decisions
are less likely to use forecasts.
  The Corning Glass Company uses a derivative form
of Delphi in its shorter range planning for developing
new products. The planning division asks its engineers
what new products will be possible in the coming year
and asks management how much money will be avail-
able for new product development. With this informa-
tion, planners develop alternative yearly "packages" or
combinations of new products that could be developed
within the funding limitations. Concurrently, Corning
develops a set of objectives for their share of the
market, long-run profits, and other company goals. A
working panel of 12 members then determines what
effect each alternative will have on the company and
on its goals. On the basis of this information, a deci-
sion is made as to which packages will be pursued.
Two articles by John C. Chambers and others at Corn-
ing describe this method of planning.36 Dalkey stated
that the Corning process is almost continuous, but he
emphasized that once the objectives of the company
are established, they are infrequently modified.
  Monsanto Chemical Company, a firm that has pio-
neered in the use of futures research for deciding prac-
tical policy, has applied the Delphi method to gen-
erate data for its top mangement to plan for the long
range. Monsanto also uses the cross-impact analysis
developed by Ted Gordon of The Futures Group,
which takes into account the effect of one event on the
probability of other events. (Delphi is used to make
the probability estimates necessary for cross-impact
analysis.) The technique also relates the effect of
selected events to Monsanto's goals. The case of Mon-

santo is one of the few in which one can document the
use of Delphi to arrive at decisions on policy.
  The largest Delphi study yet attempted was done by
the Japan Techno-Economics Society.37 A panel of
4000 engineers, scientists, and intellectual leaders in
Japan was polled on 620 subjects— to forecast for the
Japanese government future scientific and technologi-
cal developments. It was the first major attempt on the
part of an organization of government to utilize Delphi
in its long-range planning. Unfortunately, the content
was sufficiently remote from policy issues of interest to
the Japanese government and the study was so exten-
sive that the government had great difficulty using it.
  Dalkey noted that the Air Force had made little use
of the technique, whereas the Army and Navy have
used it more extensively. The CIA was interested in us-
ing it about three years ago but, to Dalkey's knowl-
edge, it has never done so extensively.
  Dalkey remarked that the largest number of appli-
cations of Delphi occurred in the field of education.
For example, Miami-Dade Community College used
Delphi to set the objectives of the school and to
measure the effectiveness of efforts to reach them.
  Abt Associates is performing research in drug
abuse. The study, sponsored by the National Institutes
of Health, uses Delphi methods to evaluate the goals of
several programs. There is also an exploratory survey
being done by the U.S. Public Health Service to review
objectives of research sponsored by the service.
  Rand also has applied Delphi to prevention and
treatment of drug abuse.38 Other Rand applications
include population estimates for cities,39 forecasting
energy demand,40 intercity transportation,41 educa-
tional innovation,42 and cable television.43
  An especially promising application of Delphi is in
determining value judgments of groups, notably in
judging the quality,of life. In a 1968 paper, Dalkey
reviewed several "armchair" surveys of human happi-
ness, satisfaction, and mental health. He then pro-
posed a research effort based on Delphi:
  The basic idea is quite straightforward, namely, to prepare a
comprehensive set of scales relevant to the quality of life; let a large,
representative sample of Americans rate themselves on these scales
via confidential interview; and employ factor analysis to summarize
the interrelations between the ratings. With any luck at all, many of
the factors derived in the analysis would be interpretable and could
replace the armchair lists with something more solid. . . .
  Since one expectation would be that the results of such a study
would be relevant to policy in the urban and domestic areas, several
blocks of scales should be allocated to issues directly involved in
these, e.g., amount of time spent in parks and places of public rec-
reation, satisfaction with neighborhood, amount of income from
welfare payments, and so on.4*
  
36

SMITHSONIAN STUDIES IN AIR AND SPACE



  By 1970 Dalkey and others had completed several
Delphi studies and had developed a model for the
quality of life (QOL) "that has been useful in sug-
gesting lines of investigation and in interpreting data.
The model includes a set of general characteristics or
qualities of the stream of events occurring to an in-
dividual that largely determine his sense of well-
being." With the model "one can estimate the poten-
tial level, but not the precise pattern, of QOL at a
given time." 45 The model was used to examine the
panel's satisfaction with different modes of transporta-
tion. This examination guided the Department of
Transportation in setting policy, thus illustrating the
usefulness of the method "to those interested in the im-
pacts of the variety of domestic public programs—
education, health, housing, transportation, welfare
and the like— upon the quality of life." 46
  In a 1973 study, Dalkey improved the QOL proce-
dure further. Respondents were asked to estimate sta-
tistical weights for those factors that they judged to be
important in determining the quality of life of an in-
dividual. Cluster analysis, a statistical technique for
grouping data, was used to aggregate the long list of
factors that were elicited. Relative importance of the
factors was measured by a Delphi exercise and by a
self-rating of overall quality of life. From the study
Dalkey concluded:
     [T]he sense of well-being enjoyed by an individual depends
upon the extent to which his experiences exemplify several basic
characteristics. Inherent in this view is the existence of trade-offs
among these characteristics; two different individuals can enjoy
about the same level of quality of life with highly different patterns
of experiential rewards. One can be comfortable, receiving a great
deal of love and affection, and living a routine existence; the other
can be living an exciting life, with a high sense of achievement, and
lonely, and each report about the same overall level of well-being.4'
  Bell Canada, a communications firm serving On-
tario and Quebec, has used Delphi extensively for
business planning. The firm has explored future
technological developments in education (e.g., com-
puterized library systems, computer-aided instruction
systems, and visual display systems such as instant
retrieval television), which suggest a market for Bell
Canada. Bell Canada conducted similar Delphi exer-
cises in medicine (e.g., multiphasic screening,
computer-assisted diagnosis, remote physiological
monitoring), in business information processing (e.g.,
management information systems, mini and small
computers, terminals and data processing), and in
home telecommunications services.48
  Delphi has been applied to the future of pharma-
ceuticals—conducted by Smith, Kline, and French 49

and by three other pharmaceutical manufacturers.50
  Two Delphi studies dealing with environmental
topics are notable. In 1972 the program of Research
Applied to National Needs of the National Science
Foundation convened a workshop on energy and the
environment to provide a background report for the
Senate Interior and Insular Affairs Committee. Its
product, "Growth Rate of Energy Demand," was pub-
lished as a Rand document.51 This report focused on
the need for energy conservation and recommended
policies ranging from education and voluntary meas-
ures to rationing and bans on consumption of energy.
  The Environmental Protection Agency used Delphi
as one of six techniques for estimating the damages
caused by air pollution in the United States. The
report commented:
  Given the dearth of air pollution dose-response information, it is
possible that the delphi method, which relies on subjective opinion
rather than objective data, will be used in a more significant way. It
seems obvious that where substantial information is missing, the
pooled judgments of experts could provide useful information to the
decisionmaker on the general magnitude of damages over a range of
pollution levels.52
Transferring Delphi
  Transfer of Delphi from its early use on military
problems to more general technological forecasting
has been accomplished largely in two ways: first, by
experimenters at Rand and elsewhere who have found
it adaptable to many other areas of interest to society;
and second, through published accounts of the meth-
od, which have been extensive. Dalkey credits the suc-
cess of the Delphi technique to its use of firm and
systematic procedures for the first time in forecasting
by judgment. The simplicity of the procedures and the
many obvious applications of the technique make it
easy to apply. Although it has been described here as
an aerospace-related technique because of its origins,
there is nothing that intrinsically limits its applicabili-
ty to air and space.
  Many applications of Delphi have been made by in-
dividuals who read or heard about it and proceeded
from written instructions in the literature. Some face-
to-face contact with the original developers has oc-
curred at professional meetings and in personal visits.
But in the main, all that is necessary to employ the
technique is a publication or two from the Rand Cor-
poration or the "methodology" section of a previous
study.
  Its flexibility and ease of application do not, how-
ever, assure that Delphi always can be used successful-
  
NUMBER 3

37



ly. As proponents and critics agree, there are many
cases where the Delphi technique has been poorly
handled and improperly applied.
A Study of How Delphi Is Used
  In 1973, the College of Business and Administration
of the University of Colorado hosted the Fourteenth
Annual Meeting of the Mountain-Plains Management
Conference. The theme of the meeting was "Manage-
ment Education: Past, Present, and Future." Among
those planning the conference there seemed to be a
consensus about the past and present state of manage-
ment education, but obtaining a prediction of future
trends presented a more difficult problem.
  The planners theorized that future trends in man-
agement education might best be predicted by leaders
in education. Not only would these persons be aware
of dominant trends, but they would be in academic or
administrative positions where their opinions also
create the trends. Since it was impossible to gather
even a small portion of these people together in one
place, the planners decided to use Delphi, in advance
of the conference, to collect their opinions and develop
a forecast of the future.
  Several of the conference planners were familiar
with the Delphi technique. One had talked with Nor-
man Dalkey at a professional meeting; one had used
the technique in a different context; and a third was
familiar with the concept from his teaching in man-
agement science, but had not used the technique. The
three planners compiled a set of leading questions
about future trends in management education. These
questions, and a cover letter explaining the purpose
and technique, were sent to the panel.
  The 60 candidates for the panel represented a varie-
ty of schools and professional points of view across the
country. Since the number of potential respondents
was large, one planner acquired a computer program
for processing the responses and for performing the
statistical analysis in later iterations.53
  Approximately 40 percent of the panelists answered
the initial request. They responded to 70 statements
about trends in management education that might de-
velop in the next 15 years by placing them in one of
seven categories from "highly unlikely" to "highly like-
ly.' The first were aggregated, refined, and cate-
gorized for future iterations. The results were then
provided on a computer-generated form. Model re-
sponses and interquartile ranges were indicated, as
well as the absolute number of responses recorded in
each category.
  
After the initial questionnaire, the procedure was
iterated twice. Those respondents who deviated from
the majority were asked for comments to support their
positions after the second iteration. General comments
from all panelists were also requested. After the third
iteration the researchers constructed a consensus state-
ment and minority statements when appropriate.
Lack of time prohibited further iterations. The results
of the forecast were mailed to the participants.
  At the conference the participants were also asked
to rate the statements generated by the second itera-
tion. Their responses were processed by computer, and
they compared their distribution of opinions with that
of the expert panel in a review session on the second
day of the conference. This process provided a good
opportunity not only to demonstrate but to utilize the
technique, thereby facilitating its future use.
  This use of the Delphi technique was judged to have
been a success by the conference program planners,
the conferees, and the panel of respondents. The con-
ference planners had been able to obtain, at a total
cost of under $200, the opinions of many of the most
prestigious management educators whose judgments
would otherwise have been difficult and expensive to
obtain. Those respondents of lesser professional
stature had in no way been intimidated by those of
greater prestige. The second and third iterations had
shown considerable convergence in the range of
responses.
  Finally, comments submitted by those deviating
from the majority showed they had thoughtfully con-
templated the experts' responses to the 70 statements.
In general, both the panel of experts and the con-
ference participants were highly complimentary of the
procedure and the data it yielded. Only time will
reveal the degree of accuracy of the forecast.
  Additionally, the program planners presented their
results to the Academy of Management at its national
meetings in 1973, thus allowing others to see their
methodology and their evaluation of the procedure.
Since that time another potential user has requested a
fuller explanation of the methodology and the com-
puter program used to process the data.
The Validity of the Method
  Although most observers agree on the significant ac-
complishments of the Delphi technique, a 1974 Rand
critique by Harold Sackman attacks Delphi as "basi-
cally an unreliable and scientifically unvalidated tech-
nique in principle and probably in practice." 54 No at-
  
38

SMITHSONIAN STUDIES IN AIR AND SPACE



tempt is made here to reconcile the opinions of Sack-
man and the founders of Delphi, but we do present
their disparate views and some comments on the rea-
sons for the disparity.
  Most managers and forecasters who have comment-
ed on Delphi seem satisfied that it is a useful tech-
nique. Delphi is normally used only when a lack of
data precludes conventional forecasting, and it pro-
vides results believed to be more trustworthy than
those obtained from alternative methods. Even though
little is provable as to why Delphi works (as Helmer
and Dalkey stress), many of its users are unconcerned
with its theoretical deficiencies. Sackman, in contrast,
applies formal tests of the social-scientific method to
Delphi and finds it wanting in statistical reliability and
adherence to accepted methodological standards. He
thus condemns its use. In Sackman's view,
The history of Delphi reveals a highly exploratory and tentative
technique that was never validated. Delphi was obviously full of
problematic issues and potentially serious flaws, and was treated
with some measure of caution and skepticism by its Rand originators
before the Gordon-Helmer study (1964) catapulted the technique
into international prominence. After that point, the shaky
hypotheses on which Delphi rested were apparently transformed into
axioms, and Delphi was promoted as an established, proven tech-
nique. . . .
  Delphi has led a protected existence for the decade it has been ac-
tively pursued. From exploratory and tentative beginnings at Rand,
it has spread from government to industry and academia, and diver-
sified from scientific and technological forecasting to policy studies
and planning, to quality of life assessment .... Droves of eminent
people and experts from all callings have lent their name, time, and
effort to hundreds of Delphi investigations. All this, and undoubted-
ly more to come."
  J. P. Martino56 has tested the accuracy of Delphi by
comparing events with earlier forecasts of Delphi
panels. Sackman scoffs at such tests as "bootstrap
validation—Delphi validating itself," and cites one of
Helmer's published Delphi studies that wrongly pre-
dicted the winner of the 1952 presidential election as
evidence that "Delphi results will often be untrust-
worthy." 57 In a 1975 interview, Sackman referred to
other cases in the Department of Defense in which
Delphi produced short-term forecasts that proved to
be inaccurate; however, he could not cite these classi-
fied studies.
  How valid is Delphi? Dalkey and Helmer recognize
the need to strengthen the theoretical bases of the
technique. Dalkey began about 1965 to conduct ex-
periments with UCLA students to test the effectiveness
of the technique and the factors promoting conver-
gence. He wrote in 1967 :
  
There are many things we do not understand as yet about the in-
formation processing going on during a Delphi exercise. Thus, we
cannot as yet determine how much of the convergence is due to three
different factors which are clearly at work: (1) social pressure, (2)
"rethinking" the problem, (3) transfer of information during feed-
back. Several exercises have been conducted that throw some light
on this.68
  Dalkey's experiments continue. In a 1974 interview
he stated that he believes some interaction is valuable
and he is trying to refashion Delphi to obtain such in-
teraction without biasing the judgment of others. He is
also experimenting with the ability of groups to judge
the relevance and the "solidity" of information (that
is, how well it has been confirmed). While persons
seem able to judge the relevance of a fact well, they are
less able to judge the solidity of a relevant statement.
Dalkey has found that the accuracy of a group in judg-
ing solidity of information depends on the group's de-
termination of its relevance. If facts are presented in
the order of the group's rating of relevance, group ac-
curacy is highest with the first fact and decreases with
each subsequent fact. Dalkey is also attempting to cre-
ate a theory of probabilistic prediction by groups
based on a scoring system for individual probabilistic
predictions.
Helmer has also called for further experimentation:
  Delphi still lacks a completely sound theoretical basis. This is due,
largely, to the fact that Delphi, by definition, is concerned with the
utilization of experts' opinions and that experts are rarely available
as experimental laboratory subjects. Delphi experience, therefore,
derives almost wholly either from studies carried out without proper
experimental controls or from controlled experiments in which stu-
dents are used as surrogate experts. It is still an open question which
of the many results obtained through this latter kind of experimen-
tation carry over to the case of real experts, and I hope that further
investigations in this area will be undertaken.
  Further solidification of the Delphi technique, based on careful
experimentation, clearly would be desirable, especially in view of
the far-reaching applications to which the method in principle lends
itself.59
  Sackman criticizes the experiments conducted thus
far:
Only relatively recently have Dalkey and some of his coworkers made
attempts to demonstrate the validity of Delphi . primarily with
almanac-type items and non-expert panelists such as college stu-
dents. These efforts, and spotty returns from a small number of
other studies mentioned in this review, provide no scientific valida-
tion of Delphi. This history of early experimentation and tardy ef-
forts to assess validity reflects a pattern of isolationism from the
mainstream of behavioral research.60
  A recent study by Van de Ven and Delbecq, pub-
lished after Sackman's critique, used rigorous statisti-
cal techniques to measure the relative effectiveness of
  
NUMBER 3

39



Delphi in comparison with two other processes for
making decisions: nominal group technique (NGT),
developed by the authors, and interacting group proc-
esses. Among other findings, the authors concluded:
. . . the degree of differences in effectiveness . . . between Delphi and
interacting groups is important and large. These differences are so
convincingly large that the practitioner should change his conven-
tional pattern of using the interacting group meeting process in fa-
vor of either NGT or Delphi techniques on applied problems . . . .61
  In place of Delphi, however, Sackman proposes the
use of rigorous techniques of administering question-
naires by trained social scientists. In an August 1975
interview, Sackman revealed that he was in the early
stages of a "feasibility study" on a successor to Delphi,
a rigorous technique that would avoid the pitfalls that
he finds in the Delphi method. If financial support
materializes, he expects to publish an outline of his
technique, "participatory polling."
  Sackman stated in the interview that he believes
many knowledgeable observers agree with his views on
Delphi. Response to his Rand report has been "gener-
ally favorable although a number of Delphi propo-
nents have taken exception. There is a strong vested
interest in supporting Delphi." He believes that the
great majority of the Rand staff is sympathetic to his
view and noted that some members of the staff had re-
cently asked to undertake a Delphi study but were dis-
couraged from doing so by staff reaction.
  Professional relationships between proponents and
critics of Delphi, while correct by academic standards,
have generated strong feelings. Norman Dalkey, when
asked about his conversations with Harold Sackman,
replied that Sackman has never contacted him directly
to explore differences in views on Delphi— an omission
that Dalkey considers significant. Sackman pointed to
a formal debate—an "adversary hearing" held at
Rand in early 1974 —between Dalkey and him in
which each presented his views before 40 to 50 of the
Rand staff. Sackman acknowledges using several of
Dalkey's comments in that debate in his Rand report.
  The second number of volume 7 of Technological
Forecasting and Social Change (1975) is devoted en-
tirely to Delphi and is largely concerned with its validi-
ty. Four articles propose new procedures or extensions
of Delphi. One article by Hill and Fowles62 echoes crit-
icisms by Sackman: that Delphi lacks recognized
standards for implementation and thus may be admin-
istered unreliably; that the evidence presented to sup-
port the validity of Delphi results is unconvincing, par-
ticularly when the date of the predicted event still lies

in the future; and that the evidence of Delphi's meth-
odological validity under test conditions cannot be
generalized to all of its uses. Finally, Hill and Fowles
criticize the utility of Delphi on the rather weak
ground that Delphi is more often used for planning
than for policymaking.
  Defenders of Delphi, however, stress its utility.
Joseph Coates comments: "Delphi is not a way of
drawing forth expert knowledge on expert's issues, but
rather is an aid to dealing with those most crucial of
contemporary societal problems: judgment and wis-
dom about the future." 63 Sam Scheele remarks that
Sackman's criticism misses the point and compares it
to a hypothetical inspection of Alexander Graham
Bell's laboratory by the Occupational Safety and
Health Administration. Scheele concludes: "The
Delphi technique can make the process of formulating
policy and making of assumptions accessible," by
bringing a kind ol participation in processes of society
to those affected.64
  In what one Delphi proponent refers to as the defin-
itive rebuttal to Sackman, Peter Goldschmidt attacks
nearly all of the bases of Sackman's criticism, accuses
him of missing the mark, and terms his criticism a
political rather than a scientific document. Acknowl-
edging that Delphi applications often have been poor-
ly chosen  or  poorly  conducted,   Goldschmidt  sum-
  Some decisions require information that cannot be derived from
knowledge, because none exists, nor from empirical study, because
this is infeasible or impractical. In these situations, either the deci-
sionmaker must rely on his own experience or on the opinions of
others who are considered to be more knowledgeable about the sub-
ject than he. The problem for the decision-maker is how to secure
such expert opinion, and more importantly, how to reconcile differ-
ences in the opinions he is offered. Delphi is a way of overcoming
this problem.65
Analysis of Significance
  Past: Rand's retrospective book reviews the accom-
plishments of Delphi:
  Much of the work of the past quarter-century was successful —
some of it singularly so—because it clarified major policy issues,
changed the attitudes and perspectives of public leaders, and influ-
enced events in important ways. Or because it resulted in new ways
of thinking about the problems of government and new methods for
dealing analytically with them. Or because it brought about a syn-
thesis of existing knowledge for application to new problems. Or
simply because it added to the common fund of what is known about
the world and how it operates.66
  
40

SMITHSONIAN STUDIES IN AIR AND SPACE



  Present: As the section on "Applications' shows,
Delphi is widely used in government and industry. Its
simplicity and economy make it attractive to small in-
dustries engaged in technological forecasting.
  Future: Lawrence H. Day of Bell Canada forecasts
a bright future for Delphi in the corporate environ-
ment. In so doing, he describes several modifications
of Delphi techniques:
  The near future should see continued rapid expansion of the
Delphi technique in business. The methodology appears to be cur-
rently reaching the "faddish" stage. Many low quality studies (which
may be mislabeled "Delphi") will be conducted. This could result in
a credibility gap with those trying to use the technique to its best ad-
vantage. If this credibility gap does occur, there may be a numerical
decline in the number of studies conducted, but a general improve-
ment in the overall quality of corporate Delphi research.
Widespread use of the methodology will result in continued rapid

modification of the original RAND design. Mini-Delphi's will be
used to develop specific forecasts or evaluate potential policy
changes. The latter area will receive further attention with the con-
tinued development of interest in technology assessment. The use of
on-line Delphi techniques will spread, especially as corporate man-
agement information systems and remote access terminals become
widespread.67 The availability of standard packages that permit any
researcher with access to an on-line Delphi system to act as a study
director will also encourage further use of the technique.68
  Delphi will become popular for certain types of market research
studies. This will probably occur more as a result of the promotional
activities of market research firms than from the conscious decision
of corporate researchers or marketing academics . . . .
  In conclusion, Delphi has a healthy future in the corporate envi-
ronment. This is a future for a whole family of Delphi inspired tech-
niques in a broad range of applications. Use of the term "Delphi" to
describe a monolithic technique has rapidly become obsolete in this
environment. This expanding family of techniques will be the prop-
erty of the market researcher, market planner, policy planner, sys-
tems researcher, etc., as well as the long term business planner.69
  
Appendix 1

Experiment I
Group 1, Group 2: Round 1, Day 1
  The experiment in which you are taking part this afternoon is one
of a series investigating human information processing. Your pri-
mary task will be to answer twenty questions of a general informa-
tion type. You will be given ten questions at a time. In addition, you
will take a standardized concept mastery test. The primary ques-
tionaire is not a quiz or an examination, nor is it a test of social in-
fluence. We are interested in the way groups use incomplete infor-
mation to arrive at factual conclusions.
  It is not expected that you will know the exact answer to any of the
questions. However for most of them you will have some general
knowledge that will enable you to make an estimate— an informed
guess—of the answer. Your are to make as good an estimate as you
can; but in any case, answer every question as best you can. You are
not required to give the answers in whole numbers—numbers with
decimal fractions are acceptable. The time allowed is 17 minutes for
each set of 10 questions, so you will have about a minute and a half
per question.
  In addition, you are asked to rate the questions with respect to the
amount of knowledge you feel you have concerning the answer. Do
this as follows: before giving any answers, look over the questions
and find the one that you feel you know the most about. Give this
question a rating of 5 in the box on the left labelled "self-rating.'
Then find the one you feel you know the least about, and give this a
rating of 1. Rate all the other questions relative to these two, using a
scale of 1, 2, 3, 4, or 5. Thus, a question about which you know al-
most as much as the one you rated 5 could also get a 5 rating. One
that you feel is roughly halfway between the one on which you are
least informed and the one on which you are best informed would be
rated 3, and so on. Notice that the rating is purely relative, and de-
pends only on how much you feel you know about the question. Do
not try to make refined estimates of these ratings, but be impres-
sionistic.
  
Do not discuss the questions or the experiment with any of the
other participants. Feel free to leave the room if necessary, and feel
free to smoke. The experimenter will be available during the session
if any questions concerning your task should arise. Notice that there
is an additional form on which you should record your answers be-
fore the questionaires are collected.
Group 1: Round 2, Day 2
  Your task this afternoon will be to reconsider your answers to the
twenty questions you worked with yesterday, and make any revisions
which, on second thought, you feel are called for. The questionaire
has a record of your answers of yesterday in the lefthand column. Do
not hurry, but rethink each question, considering whether there
were factors you might have overlooked, or computations which
might have contained numerical mistakes. However, keep in mind
that you are still being asked only for your best estimate, based on
what you know.
  There is an additional form on which you should record your
present answers before the questionaires are collected. You will save
this record of your answers for the second part of today's session.
Group 2: Round 2, Day 2
  On this round, you are to reconsider your answers, this time tak-
ing into account some information concerning the responses of the
other participants on the preceding round. You have your record of
your previous answers. In addition, the questionaire has a descrip-
tion of how the answers of all the other participants in your group
were distributed. This description is given in terms of the Median
and the Quartile interval, QJ-QJ. The median is the middle re-
sponse for that question. The quartile interval, QJ-QJ, contains the
middle 50 percent of the responses; its size gives you some indication
of how widely the responses differed from one another.
  
NUMBER 3

41



  Taking this new information into account, you may revise your
answers where you think it appropriate.
Group 1, Group 2: Round 3, Day 2
  On this round you are being asked to reconsider your estimates in
light of the changes in estimates by the group on the last round.
Your questionaire has the summary of the group's second responses,

and you have your record of your own answers. Please record your
present answers below the first set on the answer sheet.
  SOURCE B. Brown, S. Cochran, and N. Dalkey, The Delphi
Method II: Structure of Experiments, Rand Memorandum RM-
5957-PR. Santa Monica, California: The Rand Corporation, June
1969, pages 13-15.
  
Appendix 2

  The following extract is quoted from the list of ques-
tions used in Delphi experiments ordered on group ac-
curacy in Round 1.
213.	Error:     .600	True:     2.47
What was the number of volumes in the UCLA library system
as of June 30, 1967 (in millions)?
214.	Error:     .610	True:     1.63
How many millions of persons visited Yosemite National Park
during 1965?
215.	Error:     .616	True:     540
How many B-52's did the U.S. Air Force have in operation as
of June 30,1967?
216. Error:     .629	True:     1066
How many calories are there in a pint of 80-proof gin?
217. Error:     .734	True:     324,000
How many mobile homes and travel trailers were manufac-
tured in the United States in 1965?
218.	Error:     .759	True:     7.68
What was the popular vote in millions for Taft in 1908 when
he was elected President of the U.S.?
219.	Error:     .761	True:     3.27
How many nonwhite children (in millions) under 5 years of
age were there in the U.S. in 1965?
220.	Error:     .770	True:     22
What was the total number (in millions) of telephones in use
in the U.S. as of December 31, 1940?

221.	Error:     .819	True:     86
What was the number (in thousands) of active, non-federal
dentists in the U.S. in 1965?
222. Error:     .838	True:     8.65
What is the area of the USSR in millions of square miles?
223. Error:     .839	True:     3.35
What was the circulation of Playboy Magazine (in millions)
as of January 1, 1966?
224.	Error:     1.001	True:     3.60
What is the total area (in millions of square miles) for all 50
states in the U.S.?
225. Error:     1.005	True:     49,400
What was the population of New York City in 1790?
226. Error:     1.015	True:     2.18
How many million tons of raisin grapes did California prod-
uce in 1966?
227.	Error:     1.139	True:     16,000
What was the largest price ever paid by an American zoo for a
single animal?
228.	Error:     1.214	True:     520
How many Junior Colleges were there in the U.S. in 1965 with
an enrollment of at least 100 students?
  Source: B. Brown, S. Cockran, and N. Dalkey, The Delphi Meth-
od II: Structure of Experiments, Rand Memorandum RM-
5957-PR. Santa Monica, California: The Rand Corporation, June
1969, pages 96-97.
  
Notes

  1. Olaf Helmer, Social Technology (New York: Basic Books,
1966).
  2. The use of controlled feedback of group responses in Delphi is
based on a social psychological theory that the individual respond-
ent is more likely to alter his judgment and move toward the consen-
sus of the group if he holds his opinion weakly. If he has substantial
confidence in his judgment, however, he is less likely to alter it. See,
for example, Solomon Asch, "Studies of Independence and Con-
formity, I: A Minority of One Against the Unanimous Majority,"
Psychological Monographs, volume 70, number 9, whole number
416, (1956).
  3. Norman C. Dalkey, "Toward a Theory of Group Estima-
tion," in The Delphi Method: Techniques and Applications,
Harold A. Linstone and Murray A. Turoff, editors. (Reading,
Mass.. Addison-Wesley, 1975), pages239-240.
4. "A quite convincing group of studies has shown that the

group opinion is likely to be highly influenced, if not determined, by
the views of the member of the group who does the most talking, and
that there is no significant correlation between success in influencing
the group and competence in the problem being discussed" (Nor-
man C. Dalkey, "Delphi," paper presented to the Second Sympo-
sium on Long-Range Forecasting and Planning, Alamagordo, New
Mexico, 11-12 Octber 1967, Rand Publication P-3704, page 5).
5. Dalkey, "Delphi," page 3.
  6. Norman C. Dalkey, The Delphi Method: An Experimental
Study of Group Opinion (Santa Monica, California: The Rand
Corporation, June 1969), Rand Memorandum RM-5888-PR,
pages 26, 27.
7. Dalkey, "Delphi," page 4.
  8. B. Brown, S. Cochran, and N. Dalkey, The Delphi Method
II: Structure of Experiments (Santa Monica, California: The Rand
Corporation, June 1969), Rand Memorandum RM-5957-PR.
  5. 
42

SMITHSONIAN STUDIES IN AIR AND SPACE



9.	Dalkey, "Delphi," page 4.
10. Ibid.
11. Dalkey, The Delphi Method.
  12. This section draws on several sources: two Denver Research
Institute reports, Aerospace Management Techniques (1971), and
Contract Research and Development Adjuncts of Federal Agencies
(1969) ; Rand 25th Anniversary Volume (c. 1973); and personal
interview with Norman C. Dalkey, 12 August 1974.
  13. In the late 1960s the RAND Corporation dropped to lower
case the last three letters of its name. For clarity, "Rand" will refer
to the corporation and "RAND" will refer to the continuing Air
Force Project from which the corporation still draws much of its
funding.
  14.	The RAND Corporation: The First Fifteen Years (Santa
Monica, California: The RAND Corporation, November 1963),
page 2. See also Bruce L. R. Smith, The RAND Corporation (Cam-
bridge, Massachusetts: Harvard University Press, 1966), page 30.
In ensuing years, although Rand's sponsorship has become more di-
verse, the U.S. Air Force continued to fund Project RAND as a line
item in the Air Force budget. Project RAND now represents about
one-third of Rand's total research effort.
15. Dalkey interview.
  16. M. Girshick, A. Kaplan, and A. Skogstad, "The Prediction
of Social and Technological Events," Public Opinion Quarterly
(spring, 1950), pages 93-110.
  17. Norman C. Dalkey and Olaf Helmer, The Use of Experts for
the Estimation of Bombing Requirements: A Project Delphi Experi-
ment (Santa Monica, California: The Rand Corporation, 14 No-
vember 1951), Rand Memorandum RM-727.
  18. Norman C. Dalkey and Olaf Helmer, An Experimental Ap-
plication of the Delphi Method to the Use of Experts (Santa Monica,
California: The Rand Corporation, July 1962), Rand Memoran-
dum RM-727-PR (abridged).
19. Dalkey, The Delphi Method.
  20. The Rand Corporation, A Bibliography of Selected Rand
Publications: Delphi and Long-Range Forcasting (Santa Monica,
California: The Rand Corporation, January 1974), SB-1019.
  21. T. J. Gordon and O. Helmer, Report on a Long-Range
Forecasting Study (Santa Monica, California: The Rand Corpora-
tion, September 1964), Rand Publication P-2982.
22. Dalkey, The Delphi Method.
  23. Ralph Charles Lenz, Jr., Technological Forecasting
(Wright-Patterson Air Force Base, Ohio: Air Force Systems Com-
mand, Aeronautical Systems Division, June 1962), NTIS,
AD-408085.
  24.	Joseph P. Martino. (U.S. Air Force), "An Experiment with
the Delphi Procedure for Long-Range Forecasting," IEEE Trans-
actions on Engineering Management (September 1968), pages
138-144.
  25. R. Campbell, "A Methodological Study of the Utilization of
Experts in Business Forecasting," unpublished Ph.D. dissertation,
UCLA, 1966; John Vincent Murray (USAF), "The Development
and Test of a Concept and Methodology for Long-Range Forecast-
ing," unpublished Ph.D. dissertation, University of Colorado, 1967,
University Microfilms 68-12318.
  26. James R. Bright, editor, Technological Forecasting for In-
dustry and Government; Methods and Applications (Englewood
Cliffs, New Jersey: Prentice-Hall, Inc., 1968); Robert M. Campbell
and David Hitchin, "The Delphi Technique: Implementation in the
Corporate Environment," Management Services (November-
December 1968),  pages 37-42;   Robert U.  Ayres,  Technological

Forecasting and Long-Range Planning (New York, New York:
McGraw Hill, Inc., 1969) ; Harper G\ North and Donald L. Pyke,
" 'Probes' of the Technological Future," Harvard Business Review,
volume 47, number 3 (May-June 1969), pages 68-82; Gordon
Wills, David Ashton, and Bernard Taylor, editors, Technological
Forecasting and Corporate Strategy (New York, New York: Ameri-
can Elsevier, 1969); Edward B. Roberts, "Exploratory and Norma-
tive Technological Forecasting: A Critical Appraisal," Technologi-
cal Forecasting— An International Journal, volume 1, number 2
(fall 1969), pages 113-127.
  27. Olaf Helmer, Ted Gordon, Selwyn Enzer, Raul de Brigard,
and Richard Rochberg, Development of Long-Range Forecasting
Methods for Connecticut: A Summary (Middletown, Connecticut:
Institute for the Future, September 1969), IFF Report R-5.
28. Dalky, The Delphi Method, page 16.
  29. Olaf Helmer, "Foreword," The Delphi Method: Techniques
and Applications, Harold A. Linstone and Murray A. Turoff, edi-
tors (Reading, Massachusetts: Addison-Wesley, 1975), page xix.
30. Ibid.
31. North and Pyke, "Probes," page 69.
32. North and Pyke, "Probes," page 70.
33. Dalkey interview.
  34. The Futures Group developed "interview Delphi," in which
a questionnaire is replaced by a carefully prepared face-to-face in-
terview. According to Wayne Boucher (who designed and led the
first study), panelists may be uncomfortable with the inflexibility
and space limitations of questionnaires; they can say what they wish
in an interview. More important, it is possible to insure that the
questions being asked are understood and that they are the ap-
propriate questions.
35. Rand Bibliography.
  36. John C. Chambers, Satinder K. Mullick, and David A.
Goodman, "Catalytic Agent for Effective Planning," Harvard Busi-
ness Review, volume49, number 1 (January-February 1971), pages
110-119; John C. Chambers, Satinder K. Mullick, and Donald D.
Smith, "How to Choose the Right Forecasting Technique," Harvard
Business Review, volume 49, number 4, (July-August 1971), pages
45-74.
  37. Japan Techno-Economics Society, Science and Technology
Developments up to A. D. 2000 (Toyko, January 1972).
  38. L. T. Thompson, A Pilot Application of Delphi Techniques
to the Drug Field: Some Experimental Findings (Santa Monica,
California: The Rand Corporation, June 1973), Report R-1124.
  39. P. A. Morrison, Demographic Information for Cities: A
Manual for Estimating and Projecting Local Population Character-
istics (Santa Monica, California: The Rand Corporation, June
1971), Report R-618-HUD.
  40. D. N. Morris, Future Energy Demand and Its Effect on the
Environment (Santa Monica, California: The Rand Corporation
September 1972), Report R-1098-NSF: W. E. Mooz, et al, Califor-
nia's Energy Quandary (Santa Monica, California: The Rand Cor
poration, September 1972), R-1084-NSF/CSRA, R-1115-RF/
CSA, R-1116-NSF/CSA.
  41. Z. F. Lansdowne, Analysis of Intercity Transport Im-
provements: Forecasting Demand and Evaluating User Benefits
(Santa Monica, California: The Rand Corporation, May 1970),
Rand Memorandum Rm-6255-DOT.
  42. O. Helmer, The Use of the Delphi Technique in Problems
of Educational Innovations (Santa Monica, California: The Rand
Corporation, December 1966), P-3499.
43.	R. E. Park, The Growth of Cable TV and Its Probable Im-

NUMBER 3

43



pact on Over-the-Air Broadcasting (Santa Monica, California: The
Rand Corporation, December 1970), P-4526.
  44. Norman C. Dalkey, Quality of Life (Santa Monica, Califor-
nia: The Rand Corporation, March 1968), P-3805, pages 14-15.
  45. Norman C. Dalkey, Ralph Lewis, and David Snyder, Meas-
urement and Analysis of the Quality of Life: With Exploratory Illu-
strations of Applications to Career and Transportation Choices
(Santa Monica, California: The Rand Corporation, August 1970),
Memorandum RM-6228-DOT, page v.
46. Dalkey, Lewis, and Snyder, Measurement, page v.
  47. Norman C. Dalkey, "A Delphi Study of Factors Affecting
the Quality of Life," in The Delphi Method: Techniques and Ap-
plications, Harold A. Linstone' and Murray A. Turoff, editors
(Reading, Massachusetts: Addison-Wesley, 1975), page 388.
  48. Lawrence H. Day, "Delphi Research in the Corporate Envi-
ronment, " in Delphi: The Bell Canada Experience (Montreal: Bell
Canada, 1972). Reprinted in Linstone and Turoff, The Delphi
Method: Techniques and Applications, pages 191 ff.
  49. A. D. Bender, et al, "Delphi Study Examines Developments
in Medicine," Futures (June 1969), pages 289ff.
  50. George Teeling-Smith, "Medicines in the 1990s: Experi-
ence with a Delphi Forecast," Long-Range Planning (June 1971),
pages 69 ff.
  51. Deane N. Morris, Future Energy Demand and Its Effect on
the Environment (Santa Monica, California: The Rand Corpora-
tion, September 1972), R-1098-NSF.
  52. Thomas E. Waddell, The Economic Damages of Air Pollu-
tion (Washington: Washington Environmental Research Center,
Environmental Protection Agency, May 1974), EPA 600/5-74-012,
page 37.
  53. This program had been developed by Vaughn Huckselt at
the Western Interstate Commission for Higher Education, located in
Boulder, Colorado.
  54. H. Sackman, Delphi Assessment: Expert Opinion, Forecast-
ing, and Group Process (Santa Monica, California: The Rand Cor-
poration, April 1974), R-1283-PR, page vi; also Harold Sackman,
Delphi Critique (Lexington, Massachusetts: D. C. Heath and Com-
pany, 1975).
55. Sackman, Delphi Assessment, page 65.


  56. J. P. Martino, Technological Forecasting for Decision-
Making (New York: American Elsevier, 1972).
57. Sackman, Delphi Assessment, page 59.
58. Dalkey, "Delphi," page 5.
59. Helmer, "Foreword" to The Delphi Method, pagexix.
60. Sackman, Delphi Assessment, page 65.
  61. Andrew H. Van de Ven and Andre L. Delbecq, "The Effec-
tiveness of Nominal, Delphi, and Interacting Group Decision Mak-
ing Processes," Academy of Management Journal, volume 17, num-
ber 4 (December 1974), pages 605-621.
  62. Kim Quaile Hill and Jib Fowles, "The Methodological
Worth of the Delphi Forecasting Technique, "Journal of Technolog-
ical Forecasting and Social Change, volume 7, number 2 (1975),
pages 179-192.
  63. Joseph F. Coates, "In Defense of Delphi: A Review oi Delphi
Assessment, Expert Opinion, Forecasting, and Group Process by H.
Sackman," Technological Forecasting and Social Change, volume
7, number2 (1975), page 193.
  64. D. Sam Scheele, "Consumerism Comes to Delphi: Com-
ments on Delphi Assessment, Expert Opinion, Forecasting, and
Group Process by H. Sackman," Technological Forecasting and So-
cial Change, volume7, number2 (1975), page219.
  65. Peter G. Goldschmidt, "Scientific Inquiry or Political Cri-
tique? Remarks on Delphi Assessment: Expert Opinion, Forecast-
ing, and Group Process, by H. Sackman," Technological Forecast-
ing and Social Change, volume 7, number 2 (1975), page 209.
  66. Rand25th Anniversary Volume (Santa Monica, California:
The Rand Corporation, c. 1973), page vii.

   67. L. H. Day cites Murray Turoff, "Delphi Conferencing,"
Journal of Technological Forecasting and Social Change, volume 3,
number 2, pages 159-204 (1972) ; also see Charlton R. Price, "Con-
ferencing Via Computer," in Linstone and Turoff, The Delphi
Method: Techniques and Applications.
   68. L. H. Day cites A. J. Lipinski, et al, "Computer-Assisted Ex-
pert Interrogation —A Report on Current Methods Development"
presented at International Conference on Computer Communica-
tions, Washington, D.C, October 1972, pages 13ff.
69. Day, "Delphi Research," pages 191-192.
67. 
Computer Simulation and the Securities Industry
Carole R. Cristiano

EDITOR'S COMMENT
  After the Second World War, the successful military use of
mathematical simulation prompted efforts to "model" social and
economic systems as well. The following account describes one such
effort. The Rand Corporation, which developed many techniques of
simulation, studied the dynamics of the flow of financial securities in
1969. Rand was charged with finding ways to relieve backlogs of un-
fulfilled orders and to reduce the increasing frequency of failures to
complete transactions. We shall see that their mathematical model
and the conclusions from it had disappointingly little utility to the
industry, though in sum we need not label them a total failure.
  Carole Cristiano has described several factors that interviewees
believed were at fault. Executives of the American Stock Exchange
and Rand informally agreed on the study in advance of a contract.
A problem arose when the larger, more conservative New York
Stock Exchange, which played no part in early discussions, accepted
cosponsorship and looked differently upon the difficulties of the se-
curities exchange industry. The New York Exchange advocated a
stock certificate depository and instructed Rand to study it. The
Amex, however, had wished to give Rand free rein to build a model
of stock flow and then to optimize that flow by varying standing
conditions of the system. The conflict apparently was never resolved.
A third sponsor, the National Association of Securities Dealers,
played a minor role in setting the direction of the study.
  Both the contractor and the sponsors committed mistakes. Frag-
mented supervision and contrary goals of the cosponsors led to some
confusion among Rand researchers. Rand seemed unwilling to ac-
cept the role of advocate of reforms, which in fact it could not avoid,
but chose to couch its results in the language of "cost-effectiveness."
Rand was also unwilling and probably would have been unable to
include social and external economic factors in its model. Thus,
while the model highly rated a full stock certificate depository be-
cause of an excellent cost-effectiveness ratio, it could not estimate
bankers' and brokers' resistance to the depository, which would
emerge from the prospect of changes in regulations and losses in
profit to the banks.
  In addition to these errors, there existed a weakness in the Monte
Carlo technique, which was employed in the simulation. Random
variables and statistical samples produced on a computer could not
be reproduced without such a machine. Thus, step-by-step analysis
of the rationale of Rand's complex program was nearly impossible,
especially because the results of that program covered thousands of
sheets of green and white paper. Rand's study lost credibility among
Carole R. Cristiano, formerly with Denver Research Institute, Den-
ver, Colorado, now with Stephen R. Fiance and Associates, Santa
Fe, New Mexico 87501.

securities exchange officials. Like an eccentric theory published in a
professional journal, the model met no resistance nor spirited discus-
sion. According to the account, the sponsors neither comprehended
nor accepted it.
  Probably all of these problems reduce to one fundamental error:
The Rand researchers, largely unaided by their sponsors, failed to
consider the entire system of securities flow. Those components that
they excluded—the social, political, and unquantifiable
economic—exerted such a perturbing influence on the real system
that their absence in the model invalidated it. Indeed, it is an unfor-
tunate but widely acknowledged fact that such components usually
cannot be included in a quantitative model, and one is therefore
obliged to conclude that some social systems may not be simulated
on a computer. To discover limits of validity of mathematical simu-
lation, however, is in itself valuable. Cristiano's illumination of some
of these limits is the major result of her study.
  During the late 1960s and through early 1969, the
financial securities industry enjoyed general prosperity
because of rising prices and high volumes in trade. At
the same time, however, the industry was burdened by
great quantities of paperwork and by an increasing
number of failures to complete security transactions
on settlement days. Members of the securities commu-
nity had discussed many suggestions for improving this
situation, but no one had been able to adequately de-
fine the problem and evaluate proposed solutions.
  In early 1969 representatives of the American Stock
Exchange (Amex), the National Association of Secu-
rities Dealers (NASD), and the New York Stock Ex-
change (NYSE) met with representatives of the Rand
Corporation, a research organization with strong ties
to the U.S. Air Force noted in the previous study.
After considerable discussion, Rand proposed to de-
velop a mathematical model to simulate the process of
trade completion and to provide a method for evalu-
ating proposals that might reduce the level of incom-
plete transactions.
  In this paper we discuss the adaptation of mathe-
matical simulation to some problems of the securities
industry. The experiences described in the study illus-
trate some of the difficulties in transferring technical
methods between economic sectors, which arise when
  
44

NUMBER 3

45



their leaders have different goals and differing views of
the role of technology.
Description of the Technique
  Simulations generally employ simplifications of
reality—physical or mathematical "models" —that
more or less accurately portray the features of a real
system. They are particularly useful for solving prob-
lems of "systems analysis" when the complex systems
under consideration cannot be analyzed using formal
analytical methods.1 Frequently, simulations represent
mathematically the interactions of components of sys-
tems. Such mathematical models lend themselves to
computer manipulation, for computers provide the
means for rapid and thorough search of optimum
solutions.2
  Of all computerized mathematical simulations, the
Monte Carlo method, a technique that was developed
during World War II, is one of the most frequently
used. Monte Carlo methods use random variables
evaluated by an artificial sampling experiment to sim-
ulate the results of a real process. Although they have
been in use for more than three decades, during the
last 10 years these methods have found extensive use in
the fields of operations research and nuclear physics.3
Problems handled by Monte Carlo methods are of two
kinds: probabilistic and deterministic. The former are
concerned with the behavior and outcome of random
processes, while the latter show probabilities of a range
of theoretically determinate events. A Monte Carlo
program can even be used to develop an approximate
answer to such a deterministic problem as evaluating a
definite integral.4 In a probabilistic problem, the sim-
plest Monte Carlo method supplies a model with ran-
dom numbers, chosen in such a way that they simulate
the random process of the original problem. The re-
searcher then infers the desired solution or modifica-
tion of his model from the response of the model to
these numbers.
  Monte Carlo was first used on a large-scale basis in
the Manhattan Project. Researchers there simulated
the probabilistic problems of diffusion of neutrons in
fissionable material. In early development of the
Monte Carlo technique, John von Neumann and Stan-
islaw Ulam introduced methods to reduce the variance
of statistical samples. That is, they reduced the uncer-
tainty in the degree to which the behavior of a random
sample differs from the behavior of the whole.
Variance-reducing vastly improved Monte Carlo by
lessening the need for large numbers of observations

while maintaining the reliability and accuracy of the
method.
  Enrico Fermi, von Neumann, and Ulam popular-
ized the use of Monte Carlo technique among physi-
cists in postwar years. While it is applicable only to
some special problems, years of research and high-
speed computers have made it a practical tool in many
kinds of analytic simulation.5
Simulation at Rand
  Two years after incorporation, Rand began in 1950
to diversify its sponsorship. In addition to Air Force
R&D activities, Rand undertook work for the Atomic
Energy Commission. In 1959 the Advanced Research
Projects Agency of the Department of Defense and the
National Aeronautics and Space Administration be-
came research sponsors. In the early 1960s the Nation-
al Science Foundation, the National Institutes of
Health, and the Agency for International Develop-
ment in the Department of State were added to Rand's
list of government clients.
  Limited research on civil issues had begun in the
mid-1950s, but it was rapidly expanded in the 1967-
1968 period, and by 1969 Rand's funding in civil re-
search constituted almost one-third of its annual
budget.6 In its diversified work, staff members often
used techniques developed or refined in Rand's aero-
space research. Simulation, operations research, and
systems analysis were employed to deal with problems
of national defense and management of air craft
maintenance. The Monte Carlo method was one of
several kinds of simulation employed to solve complex
problems of Air Force management.
  In the 1960s the Air Force performed simulation
studies under Project RAND that utilized Monte Carlo
techniques employed later in Rand's analysis of the fi-
nancial securities industry. Among these were numer-
ous studies of proposed systems for maintaining Air
Force bases.7 Two persons involved in the stock ex-
change study (John Lu and James Peterson) were also
participants in these Air Force studies. Robert
Petruschell, principal investigator for the securities
industry simulation, and several of his colleagues had
been peripherally involved in them and had partici-
pated in other studies using Monte Carlo methods."
Needs of the Financial Securities Industry
  In 1969 it became apparent that the securities
community  had  several  complex  problems,   which
  
46

SMITHSONIAN STUDIES IN AIR AND SPACE



needed independent evaluation. Since the beginning
of the 1960s, both major exchanges (NYSE and
Amex) had experienced difficulty in handling the
ever-increasing volume of securities transactions. A
large (and unforeseen) increase in volume created
even more serious problems for the industry in early
1969. The orders caused backlogs of paperwork in
broker-dealer operations and an increasing number of
incomplete transactions. Unprecedented numbers of
customer complaints drew critical attention from the
exchanges, the National Association of Securities
Dealers, the Securities and Exchange Commission,
and from the Congress.
  The Senate Committee on Banking, Housing, and
Urban Affairs considered scheduling hearings to find
solutions to these problems, which many thought
would become more severe. Members of the industry
wished to develop their own solutions to these prob-
lems prior to the hearings.9 Several approaches were
taken to obtain these solutions: NYSE retained John
Cunningham as executive vice president of the ex-
change to implement a NYSE plan for a central certif-
icate depository; North American Rockwell was con-
tracted to perform a systems analysis of the paper flow
and to recommend ways to expedite this flow; Ernst
and Ernst was retained to state the problems of and
find the solutions to maintaining accurate day-to-day
accounting records; and Amex approached Rand to
study and to recommend long-term improvements
within the exchange process.10
Origins of the Rand Study
  Rand became involved largely because of a friend-
ship between Ralph Saul, president of the American
Stock Exchange and former director of Trading and
Marketing for the Securities and Exchange Commis-
sion, and David Novick of Rand. Informally Novick
and Saul had discussed some problems of the securities
industry, especially the greater frequency of incom-
plete transactions. They agreed that if the industry
were to solve its problems it would need a thorough
analysis by an independent organization with experi-
ence in solving complex problems—an organization
like Rand. Saul and Novick began to discuss the possi-
bility of letting a contract to Rand. At that time,
Amex could not describe the problems of the industry
in a form usable to Rand, but Rand was hired to per-
form a two-month exploratory study to observe the
problems and to define what, if anything, its research-
ers could do to assist the industry in solving them. Be-

cause Rand has been restricted by its charter to under-
taking research in the public interest, it was prohibited
from working for Amex alone. Therefore, a contract
to Rand was sponsored jointly by Amex, the New York
Stock Exchange, and the National Association of Se-
curities Dealers. While its "public" nature was thereby
assured, the project became much more difficult
because of its multiplicity of sponsor interests.lI
  Rand concluded the initial study by proposing to
analyze the entire structure and the transaction proc-
ess of the financial securities industry. By developing a
Monte Carlo simulation model they would study the
trade completion problem. The model would generate
data for daily trading activity of the exchanges, un-
completed trades, and stock loan practices. It would
simulate major characteristics of the trade completion
process, such as stock delivery priorities, stock borrow-
ing and segregating policies, stock clearing processing
rules, and others. All processes involved in the flow of
stock certificates essential to settling trades on any ex-
change or in the over-the-counter (OTC) market were
to be modeled.
  The advantages of using such a technique were
summarized by Rand:
  1. The real system continued to function without
disruption during the experiment.
  2. By eliminating processes of the real system that
are extraneous to the problem being studied, the effect
of proposed changes on the process under study can be
more clearly observed.
  3. A large number and range of changes can be
examined quickly and at low cost.
  4. Proposed changes can be studied singly or in a
variety of combinations to determine their effect on
the system and on each other.12
  Although there were differing perceptions of the
scope of Rand's research, the proposal was accepted,
and a $1.2 million contract among the three parties
was negotiated. While Rand and Amex interviewees
recalled in 1974 the immediacy of the problems and
the need for solutions that could be easily imple-
mented, an Amex press release gives some credence to
the NYSE perception of a more long-term nature of
the work.
  Rand staff members found it necessary first to study
the securities exchange process that they had proposed
to model. This task was difficult for a number of rea-
sons. First among them were problems in establishing
a working relationship with staff members of NYSE
and NASD. While the top management of Amex was
enthusiastic about the study, NYSE management was
  
NUMBER 3

47



willing to do little more than fund the study. Former
staff members of NYSE state that management had al-
ready determined its solution to the problems of the
exchanges—the establishment of a certificate deposi-
tory. The Rand study was considered superfluous; it
was an "insurance policy," a hedge against coming
political problems with the Securities and Exchange
Commission (SEC) and members of Congress.13
NASD was also involved in the study but saw its role
primarily as an aid to OTC dealers, touting NASDAQ
(a computerized automated trading system for OTC
securities) , and otherwise staying out of the way.
Conducting the Study
  Because of the lack of enthusiasm on the part of the
top executives of NYSE and NASD, their staffs did not
participate fully in the development of the project.
Robert Petruschell, principal investigator of the Rand
team, estimated that the team spent considerably
more time with the Amex staff alone than they did
with the two other staffs combined. He feels this was
due primarily to the fact that the study had been en-
thusiastically endorsed by Ralph Saul, who urged his
staff to cooperate fully with Rand.1* Former staff
members of NYSE agree that the study was initially an
"Amex baby," and they attribute their continuing lack
of enthusiasm to their impression that the study be-
came even more of an Amex concern as it progressed.
John Alexander, then employed at NYSE, states that
the Rand study team held only a few meetings with
him. The Rand researchers preferred to work most of
the time in California, making infrequent visits to New
York to consult primarily with Amex staff members.
When Alexander did meet with them, Rand research-
ers did not seem receptive to his suggestions and his
offers of NYSE automated sources of data.15 Similar
experiences were reported by other NYSE staff mem-
bers and, as a result, their enthusiasm for the study
waned. It is apparent that in the close-knit, informally
communicating securities community both Rand re-
searchers and middle-level managers sensed the top
administrators' enthusiasm or lack thereof and reacted
accordingly.
  Rand researchers spent a substantial amount of
time and project money familiarizing themselves with
securities exchanges. The Rand report of December
1970 described the major features of their model of
this process. Some, such as the transfer agent, were
represented at a high level of aggregation; others,
such as back-office activities, were represented in more

detail. The model simulated the flow of stocks origi-
nating from trading in a single issue through the back
office of a sample of firms representing the industry.
More specifically, the model was developed to perform
the following functions:
1.	Simulate daily trading activities.
a.	Generate the shares traded per day for a given issue.
b.	Allocate these shares to firms and customer types in the
model according to their respective portions of the
market.
u.      Simulate errors due to "uncompares." 16
2.	Simulate resulting back-office operations.
a.	Determine security due-ins and due-outs.
b.	Determine segregation requirements.
c.	Borrow and lend stock.
d.	Receive and deliver certificates.
Simulate denomination problems.
Simulate DK problems."
e.	Keep an aggregate version of a firm's stock records.
3.	Simulate the clearinghouse operations.
a.	Net each firm's trade.
b.	Allocate net positions.
4. Simulate delay times of transfer operations.
5. Record the levels of various kinds of failures to complete
deliveries.
6. Estimate the financial implication of failures for each firm.
7. Calculate operating statistics for selected brokerage firms and
throughout the industry.18
  After the basic model was created, the researchers
established a benchmark case to represent the existing
system of trading on the NYSE at a volume of about 20
million shares per day and at an average price of $35
per share. The model was used to generate "fails to de-
liver" in the benchmark case and the cost incurred by
brokers due to these fails was calculated. Then several
alternatives to the existing system of trades were tested
by introducing one or more changes. The costs of the
resulting fails were added to the costs of introducing
the changes, and the sums were compared to the cost
in the benchmark case. In the analysis of alternative
clearing systems for over-the-counter stocks, a similar
procedure was followed: the cost to the members of
the securities community of the existing system of
clearing was compared to the costs of two alternative
clearing systems.19
  The Rand team had conceived the model to be a re-
usable tool for the securities industry. They felt the
model was sufficiently well designed to be of assistance
in solving new problems as well as those they had ad-
dressed. Amex staff members agreed. NYSE staff,
however, did not feel that the Rand staff adequately
demonstrated that the model was an accurate repre-
sentation of the exchange process. In addition to their
  
48

SMITHSONIAN STUDIES IN AIR AND SPACE



skepticism about the model itself, NYSE staff mem-
bers were disappointed with the alternatives that Rand
chose to analyze. Rand had focused on the probable
effects of immediate solutions to current problems. It
was only at the insistence of NYSE that Rand consid-
ered any long-range solutions, specifically, the estab-
lishment of a full depository.20
  Rand staff members did not feel that their charge
was to make recommendations but merely to indicate
probable results of variously changing the securities
exchange process. Yet many interpreted several of
Rand's alternatives that were obviously cost effective as
recommendations.21
  Alternatives shown to be particularly cost effective
were as follows:
  1. A change in priorities of delivery of stock certifi-
cates, the most important aspect of which was giving
first delivery to brokers rather than to institutional
customers.
  2. The installation of high-speed computers to re-
duce time of transferring stocks.
  3. The implementation of a system to require
member firms to make stock loans to cover "fails to de-
liver" by other brokers, thereby lowering the rate of
broker-to-broker fails.
  4. The initiation of partial delivery rules, installa-
tion of an interbank wire system, continuous monitor-
ing of stock inventories, and the implementation of a
"Floor Derived Clearance System" to reduce broker-
to-broker fails.
  5. The initiation of a full-stock certificate deposi-
tory, incorporating the first four changes.22
  Petruschell presented the technical results of Rand's
research and the effects of the suggested changes to the
sponsoring organizations and other members of the se-
curities community. He was confronted with no serious
argument. He felt that the results were sound and the
considerable value of a number of the changes war-
ranted that some be introduced. The researchers also
felt that the model would continue to be used.23
Implementation of Results
  In interviews conducted in September 1974 and July
1975, several individuals in the securities industry who
observed the Rand study and its consequences indi-
cated that some of the changes were implemented.
  High-speed computers have been installed in major
banks and in the exchanges. The procurement of these
computers and their use in a program to reduce trans-

fer times was directed by a Banking and Securities In-
dustry Committee (BASIC), which set time standards
for transfers. These transfers could be met only with
computers.24
  Amex and BASIC rules now require dealers to fol-
low uniform procedures for partial deliveries. This
decision was influenced in part by the Rand study.25
  A depository for stock certificates, acknowledged by
staff members of both exchanges to be the best long-
term solution, has been implemented. Tom Mclner-
ney, president of the Financial Services Division of the
Automatic Data Processing Corporation and former
vice president in charge of Operations Planning at the
Amex, stated that the depository has revolutionized
processing in the securities industry and has far ex-
ceeded the expectations of the most optimistic mem-
bers of BASIC. Participation has increased, and part-
nerships with regional depositories have been formed.
Mclnerney stated that implementing the depository
was "the single biggest operational step the securities
industry has taken in 20 years to solve its problems,
cutting certificate movements by 75 percent." Severe
problems in the early stages, according to Petruschell,
were due in large part to a doubling of paperwork in a
two-phase transition period. The firm of Arthur
Young and Company was retained by NYSE to analyze
the clerical problems in late 1970-1971. In the end
they solved many of those problems. Mclnerney feels
that the Rand study helped to advance the depository
and gave credibility to its continued development,
despite the initial problems. However, he and
Petruschell agreed with John Cunningham that the
depository was timely and that it would have been im-
plemented without the Rand study.26 As is further ex-
plained below, the present depository is not the "full"
depository recommended by Rand. Hence its ability to
reduce the number of incomplete transactions is less
than Rand predicted.
  There were several alternatives that were not im-
plemented for a variety of reasons. Among them, there
were no changes in the priorities of delivery of stock
certificates. Giving first priority to brokers would have
eliminated more fails by eliminating chain reactions of
nondelivery from broker to broker. The securities sys-
tem as a whole would have benefited, but rarely would
a broker give priority, for the good of the system, to
other brokers before satisfying the needs of his cus-
tomers. Brokers sensed that individually, in the short
run, they would not benefit from the revised system, so
they did not initiate the change.27
  
NUMBER 3

49



  The industry did not adopt a system for mandatory
loaning to brokers of stock to cover fails. Many brokers
(especially in the larger firms) preferred to control an
inventory of stock certificates or to loan to other brok-
ers at their discretion rather than on demand.28
  The Floor Derived Clearance (FDC) System has not
been implemented. This was due in part to the fact
that during the final phases of the study, there was a
drop in trading, and the urgency of the crisis disap-
peared. Paul Kolton, former executive vice president
and now chairman of the board of Amex, stated that
the need for such a system was further reduced by the
establishment in 1972 of the Securities Industry
Automation Corporation (SIAC), a subsidiary of both
exchanges. It focuses on the problems of initial trans-
actions rather than on those in back offices.29 SIAC
differs from Rand's FDC system in that SIAC depends
on brokers to submit lists of all transactions, while
FDC would have recorded the transaction when it was
made on the floor of the exchange. Harry D. Todd,
formerly of Amex, stated that FDC would have been
much more efficient. He believes that, as exchange
needs become more complex and computer technol-
ogy becomes more sophisticated, its future implemen-
tation is almost inevitable. John Cunningham agreed,
while Alexander and Mclnerney felt that the goals of
FDC can be reached more efficiently through other
systems. All agree that SIAC works well enough to
have discouraged any major campaign to change the
system immediately.30
  Movement of certificates has not been halted. Some
banks and brokerage firms were legally required to
maintain the certificates in their vaults or within their
states. They were continually subject to audits, and
banks in particular were uninterested in lobbying to
change state legislation. The NYSE had sponsored a
study of the Uniform Commercial codes of states and
had lobbied to modify these laws, but by the late 1960s
not every state had made the required changes. Total
immobilization was not immediately possible.31
  Additionally, all involved agreed that many smaller
stockholders throughout the country would balk at al-
lowing brokers to hold their stock certificates in a de-
pository in New York. The securities industry is con-
servative, and its members may have resisted the man-
datory immobilization of certificates because it consti-
tuted a major change in accepted procedures. While
decision makers did not reject the idea of a depository
completely, they did reject one of its major tenets— the
retention of the majority of certificates in one place.32

Perceptions of and Reactions to the Study
  Adopting the Rand approach would have led to a
substantial increase in the number of regulations to
ensure that the new procedures would be followed.
Such an increase would inject unwanted rigidity into
the exchange process and would allow individual bro-
kers less flexibility. For this reason, various members
of the securities community were highly resistant to the
proposed changes.
  For example, brokers agreed that a priority delivery
system existed informally, but they rejected being
forced to recognize a rigid structure of priorities on the
authority of Rand's analytical model. Furthermore,
influential individuals within the industry could better
advance their positions with a looser structure. There-
fore, many large firms resisted this major change.
Some of these firms had already introduced the new
practices in their organizations, and they had thereby
gained competitive advantage. Required changes
would benefit the securities community overall, but
they would be detrimental to the hierarchy of firms.33
  Banks that transfer stock also profited from ineffi-
ciency. These influential banks serve as agents for in-
stitutional customers and, in New York, as agents for
other banks throughout the country. With the existing
delivery system, banks could hold a larger number of
securities than with the proposed changes. These secu-
rities were usable, interest-free holdings.
  As an example, consider a person in Los Angeles
who requests his bank to purchase 100 shares of a com-
pany. The Los Angeles bank will (a) either require
cash from the customer or will verify that he has suffi-
cient funds in his bank account to cover the purchase
of the stock, and (b) then wire its New York agent
bank to purchase the stock, but will not include in-
structions for disposition of the purchased stock. The
New York bank will instruct its broker to purchase the
stock as soon as possible. When the broker delivers the
stock to the bank, the instructions from the Los
Angeles bank have not yet arrived— the receiving win-
dow has no knowledge of the transaction. The broker
must then borrow funds from the bank to cover the
cost of the stock that he has purchased. The bank in
Los Angeles earns interest by loaning funds deposited
by the original customer, while the New York bank
earns interest on funds it would otherwise have to pay
to brokers on the receipt of the stock.
  Multiplied by the number of such incomplete trans-
actions, interest earned is substantial for the larger
  
50

SMITHSONIAN STUDIES IN AIR AND SPACE



banks. Therefore, there was disincentive to smooth out
the system. Transfer agents (generally banks) were
unautomated and slow, but "worked all right" and
were not enthusiastic about granting favors to brokers,
such as rushing transfers and making change, especial-
ly at their own expense.34
  There were less tangible results of the study. Kolton
stated that the study brought new disciplines to the se-
curities industry, giving it a different perspective on its
problems.35 Despite the intrinsic limits of a mathe-
matical model that simulates human behavior, Mc-
lnerney agreed that the industry benefited from expo-
sure to simulation by modeling. Mclnerney felt that
the strongest point of the study was its emphasis on the
need for banks to change their practices of securities
trading. The study reaffirmed convictions of several
executives and acted as a catalyst for change.36
  Those at Amex who were involved in the study
found it to be valuable. Led by Mclnerney and Todd,
staff members worked continuously with the Rand re-
searchers early in the study and were committed to
utilizing the model after it was finished. Later Amex
made limited use of the system and found it helpful.
Its use, however, was constrained both by the lack of
enthusiasm that NYSE management displayed and
practical problems in using the model. The program
was designed to be run on an IBM 360/50 or larger
computer, one of which was available only at the New
York Stock Exchange. Mclnerney stated that an an-
nual expenditure of $100,000 per year would have suf-
ficed for three to five persons to modify and test the
model on a continuous basis.37 Alexander disagreed.
He doubted its validity and felt the annual expendi-
ture would have come closer to $300,000 or
$400,OOO.38 It was an expense NYSE did not wish to in-
cur. Kolton noted that Amex has presented several
strong arguments to NYSE to solicit their continuing
cooperation but never obtained a commitment.39
  Although never stated, it became apparent that
some results of the study were contrary to NYSE pol-
icy. Former Amex Assistant Director Todd noted that
the attitude of NYSE management became negative
after Rand researchers asserted that the particular cer-
tificate depository endorsed by NYSE was not essen-
tial. Other changes, taken together, could have solved
many industry problems sooner. Even a "full deposi-
tory" would be cost effective but was not required. Be-
cause NYSE was committed to the depository, "there
was little enthusiasm for supporting an analysis that
would continue to undermine previous decisions."40
Mark Guarino, formerly an employee of NYSE, also

implied that this was a major problem. While he
accepted the logic of the model, he dismissed the com-
bined alternatives as "unrealistic." The model de-
picted results of certain actions, but it could not pro-
duce a detailed procedure to bring about the
changes.41 Cunningham explained that without diffi-
cult coordination of the actions of the banking
community, the Federal Trade Commission, and the
Securities and Exchange Commission, any changes in
exchange policy would be ineffective.42
  Rand had not been charged with developing de-
tailed recommendations to implement, nor was any-
one in NYSE assigned to study even the feasibility of
implementing the proposed changes. Neither Rand
nor Amex could adequately respond to charges that
the results were unrealistic. They could only reiterate
their request that NYSE continue to support the model
and to assist in the funding of studies of the feasibility
of implementation. By this time, however, delivery
problems had lessened, NYSE was unhappy with rec-
ommendations contrary to policy, and their response
to the request was unfavorable. Amex felt that for one
exchange to attempt to use the model would be point-
less, and it fell into disuse. The goal of Rand that it be
a useful tool in planning operations of the securities
community was never met.
Analysis of Significance
  The Rand simulation model had little impact on the
operation of the securities industry, but its problems il-
lustrate the complexity of transferring technological
expertise.
  Mathematical simulation is one of the most visible
computer techniques refined by the aerospace indus-
try. It has great appeal in the abstract, but few persons
outside engineering and science understand its range
of applications. To be able to accept and to use simu-
lation in less technical problems, one must first
understand it.
  In the aerospace industry, simulation was a familiar
engineering and managerial technique. Aerospace
managers, who frequently had technical backgrounds,
understood the principles of modeling and had faith in
the results. On the other hand, many of the managers
in the securities community of 1970 understood the
principles of simulation but not the details. In retro-
spect, this lack of understanding contributed to the
skepticism of NYSE managers and the inability of
Amex to defend the model.
In several ways Rand staff approached the securities

NUMBER 3

51



assignment as if it were a defense contract: the cited
interviews suggest that Rand researchers did not take
adequate steps to acquaint the staffs of the exchanges
with their approach, nor did they report enough to
their sponsors. "We asked them for progress reports,"
said Alexander, "and they acted surprised." 43 Most
importantly, they spent the majority of their time de-
veloping the model rather than testing and demon-
strating its utility to their sponsors. Tom Mclnerney
admitted that even he did not understand the model
fully, and he doubted whether anyone other than
Harry Todd had an adequate grasp of its potential
utility.44 This lack of understanding was compounded
by the use of the expansive Monte Carlo program.
Petruschell admitted that when one changed the
parameters of the model, it was very difficult to de-
termine exactly how the results, which filled a 6- by
18-foot table with a foot-depth of computer printouts,
had been achieved.45
  The incredibility of the model was exacerbated by
other factors. One problem inherent to simulation was
magnified in its use in the securities exchange indus-
try. Perhaps 20 percent of technical problems, such as
those involved in the aerospace industry, may be quali-
tative or "social" and 80 percent may be quantitative
and easily modeled.   In the securities industry,   the

social elements embrace a substantially higher percen-
tage of the problem. While the simulation of the tech-
nical elements of securities exchange may have been as
valid as in a technical problem, their smaller propor-
tion of the entire system brought the whole methodol-
ogy into question.46
  In an aerospace company, the goals of individuals
coincide with the group goal, which might be, for ex-
ample, an aircraft that flies. In the securities commu-
nity, the goals of individuals are often different from
those of the researchers who are seeking to increase ef-
ficiency of the entire exchange system. While the
model defined goals of the industry as a whole, it
threatened individual participants. But the model
threatened group goals as well when its results con-
flicted with a standing policy.47
  While simulation provided an objective look at the
industry and catalyzed some changes, it was not used
as a tool to manage the industry. Those who appre-
ciated its potential were not in a position to utilize it,
and top management was not committed to establish-
ing an optimum process of securities exchange. When
one attempts to apply objective criteria to solve social
and political problems, says McCormack of the Securi-
ties Industry Automation Cooperation, "you won't
make it with a simple number crunch."48
  
Notes

   1. Much oi trus material is drawn from the following source:
Martin D. Robbins, John A. Kelley, and Linda Elliott, Mission-
Oriented R&D and the Advancement of Technology: The Impact
of NASA Contributions (Denver: Denver Research Institute, 1972).
   2. J. Gordon Milliken and Edward J. Morrison, Aerospace
Management Techniques: Commercial and Governmental Applica-
tions (Denver: Denver Research Institute, 1971), page 68.
   3. J. M. Hammersley and D. C. Handscomb, Monte Carlo
Methods (New York: John Wiley and Sons, Inc., 1964), pages 1-4.
   4. Statement of Peter Warren, associate professor, Department
of Mathematics, Denver University, Colorado, 12 June 1975.
5. Hamersley and Handscomb, page 4.
   6. Rand 25th Anniversary Volume (Santa Monica, California:
The Rand Corporation, 1974), pages iv-vii.
   7. R. R. Fisher, et al., L-COM—A Logistics Composite Model
for Simulating Air Base Operations—An Overall View (May 1968) ;
T. C. Smith, SAMSOM— Support Availability Multi-Systems Oper-
ations Model, (June 1964) ; T. C. Smith, et al., A User's Manual for
SAMSOM II (November 1967); V. J. Voosen, Planet- (Planned
Logistics Analysis and Evaluation Technique) Simulation and Eval-
uation of Logistics System, (September 1965).
  8. Telephone interview, Robert Petruschell, The Rand Corpo-
ration, Santa Monica, California, 12 November 1974.
  9. Interview, John Alexander, First National City Bank, New
York, 30 June 1975.


  10. Interview, Tom Mclnerney, Automatic Data Processing
Corporation, New York, New York, 1 July 1975.
  11. Interview, Robert Petruschell, The Rand Corporation, San-
ta Monica, California, 12 August 1974.
  12. R. L. Petruschell, et al., Reducing Costs of Stock Transac-
tions: A Study of Alternative Trade Completion Systems, volume I
(Santa Monica, California: The Rand Corporation, 1970), page 3.
  13. Interview, John Cunningham, Arthur Young and Company,
New York, New York, 1 July 1975.
14. Petruschell interview, 12 August 1974.
15. Alexander interview.
  16. A trade is uncompared if the firms involved in the trade do
not agree on the terms.
  17. A DK occurs when stock is presented for delivery and the ex-
pected receiver claims no knowledge of the transaction. The receiver
says, "I don't Anow the trade."
18. Petruschell, et al., volume III, pages 2-3.
19. Petruschell, et al., volume III, page 3.
20. Alexander interview.
21. Cunningham interview.
22. Petruschell, et al., volume I, pages 5-10.
23. Petruschell interview, 12 August 1974.
  24. Telephone interview, Tom Mclnerney, Automatic Data
Processing Corporation, New York, New York, 15 September 1974.
25. Ibid.
10. 
52

SMITHSONIAN STUDIES IN AIR AND SPACE



26. Cunningham interview.
27. Mclnerney interview, 1 July 1975.
  28. Telephone interview, Mark Guarino, Securities Industry
Automation Corporation, New York, New York, 25 September
1974.
  29. Telephone interview, Paul Kolton, chairman of the board of
directors, American Stock Exchange, New York City, 17 September
1974.
  30. Telephone interview, Harry Todd, Mountain Bell Tele-
phone, Denver, Colorado, 18 September 1974.
31. Alexander interview.
32. Mclnerney interview, 1 July 1975.
33. Alexander interview.
34. Petruschell interview, 12 August 1974.
35. Kolton interview.


36. Mclnerney interview, 15 September 1974.
37. Ibid.
38. Alexander interview.
39. Kolton interview.
40. Todd interview.
41. Guarino interview.
42. Cunningham interview.
43. Alexander interview.
44. Mclnerney interview, 1 July 1975.
45. Petruschell interview, 12 November 1974.
  46. Interview, Edward J. Morrison, professor of Management
and Organization, University of Colorado, 3 October 1974.
47. Todd interview.
  48.	Telephone interview, James McCormack, Securities In-
dustry	Automation Corporation, New York, New York, 23
September 1974.
 
Systems Analysis in the California Studies of Public Services
Carole R. Cristiano

EDITOR'S COMMENT
  Carole Cristiano's review of the "California Studies" documents
how the techniques of systems analysis were used in trying to solve
four broad social "problems" in California. It suggests that systems
analysis often may be unsuitable to define or solve such problems,
which encompass a diversity of human and technological interac-
tions infinitely more complex than technical developments alone. As
more deficiencies in applying the techniques are disclosed in the
case, the reader may find his sympathies approaching those of Ida
Hoos, who declared: "... the very characteristics which distinguish
social from other species of systems render them resistant to treat-
ment . . . ." There seems to be a principle that underlay all those lit-
tle things that went wrong, from shoddy workmanship to political
naivete. Perhaps it was the reluctance of humans to be fit into ana-
lytical cubbyholes.
  After reading Cristiano's account, we must judge the "California
Studies" a failure. While one might argue that the technical results
of the studies were positive, they were not implemented and there is
no chance that they will be. Cristiano, building on an earlier report
of her colleague John Gilmore, has identified several reasons for this
lack of consequence. The four contractors might have encompassed
political forces within their systems, for these dominated the systems
in the end. The monitors of the contracts might have focused their
problems better at the outset and held their contractors to rigid
standards, for the lack of criteria apparently contributed to loosely
reasoned arguments. Those who initiated the contracts might have
"institutionalized" the studies by introducing them in the California
legislature and thus might have gained legal standing for them.
  Perhaps the most important lesson of this case, however, is that
the strength of systems analysis derives as much from quantitative,
detailed descriptions of components as it does from encompassing
the totality and diversity of those components. Ironically, one must
narrow the analysis, study the minutiae, before one can generalize.
Such a method appears bound to encounter problems when it is ap-
plied to a state of society whose detailed workings are not even
qualitatively understood.
Introduction
  In 1964, the state of California encountered increas-
ingly serious economic and social problems.   These
Carole R. Christiana, formerly with Denver Research Institute,
Denver, Colorado, now with Stephen R. Fiance and Associates, San-
ta Fe, Ne w Mexico 87501.

problems, apparent throughout the country, were
more severe in California, in large part because of
rapid population expansion. The state had moved
from a position as the fifth most populous state prior
to World War II to second place by the start of the
Korean conflict, and to first place during the mid-
1960s. Among the most serious problems were urban
congestion and shortcomings in the transportation
system, rising educational needs, environmental pollu-
tion, and a higher-than-national average level of
unemployment.1
  The last problem threatened to worsen because of
impending cutbacks in defense and space contracts.
More than one-third of the state's manufacturing
workers were employed by the aerospace industry,
which was largely dependent upon federal funding. In
1964, a year in which California's unemployment was
6.0 percent, reductions in federal aerospace contract-
ing added a further burden to the state.2
  Aerospace employees presented a different kind of
unemployment problem than did workers in other in-
dustries. Employees of aerospace and defense contrac-
tors were highly skilled and educated. They provided a
reservoir of talent generally presumed competent to
study almost any complex problem and to design sys-
tems to cope successfully with it.
  This situation created an opportunity to match
aerospace talents and urban problems on an experi-
mental basis. After informal exploration among aero-
space firms by his aides, Governor Edmund G. Brown
announced that he would invite these firms to bid on
studies to create urban systems in four critical areas.
The four systems were identified as follows: (1) an in-
tegrated transportation network within the state; (2)
a new system to handle California's criminally and
mentally ill; (3) a system of collecting information
upon which government and industry could base deci-
sions for years ahead; and (4) a system for managing
the wastes discharged into the air, soil, and water.
  These studies came to be known as the "California
Studies": applications of techniques of aerospace sys-
  
53

54

SMITHSONIAN STUDIES IN AIR AND SPACE



terns analysis and aerospace personnel to domestic
problems.3 This chapter will introduce the techniques
of systems analysis in the aerospace industry and will
document their applications to the urban problems of
California.
Description of Technique
  E. S. Quade of the Rand Corporation has com-
pleted an authoritative research on the application of
systems analysis to the defense industry. He has de-
fined the process as
an analytic study designed to help a decisionmaker identify a pre-
ferred choice among possible alternatives. It is characterized by a
systematic and rational approach, with assumptions made explicit,
objectives and criteria clearly defined, and alternative courses of ac-
tion compared in the light of their possible consequences. An effort
is made to use quantitative methods, but computers are not essen-
tial. What is essential is a model that enables expert intuition and
judgment to be applied efficiently. The method provides its answers
by processes that are accessible to critical examination, capable of
duplication by others, and more or less readily modified as new in-
formation becomes available.4
  In short, "systems analysis" is the formal examina-
tion of alternative designs of a system, a tool in a proc-
ess of rational decision making. This is not mecha-
nized decision making; rather it is a method of supply-
ing helpful information to the decision maker. The
actual decisions reflect experience, judgment, and
values. The information may be transmitted as (1) a
"best" solution, recommended after examining alter-
natives and accompanied by supporting detail and jus-
tification of that solution, or (2) alternatives
presented to the decision maker with no recommenda-
tions, but with evaluation of the inputs and outputs—
the cost and effectiveness—associated with each alter-
native.5
  The "systems approach" is a process that involves
identifying a problem or threat, learning and describ-
ing its environment, and then defining what must be
achieved to counter the threat or solve the problem.
Alternative plans must be considered, and the most at-
tractive one is then elaborated into the design of a
system. The "most attractive" method or system may
be selected by rigid "cost-effectiveness" analysis, by in-
tuition and judgment, or usually by a combination of
the two. The choice must consider the environment in
which the system will work and include deliberate cri-
teria for choosing among the alternative methods.
  While the systems approach usually begins with sys-
tems analysis, it can be extended to design, engineer-

ing, and implementation. Thus, the systems approach
can, in theory, be applied to all aspects of solving
problems.6
History of the Development
  Systems analysis emerged in the early 1940s from
solving military problems in which intuitive decisions
were often too costly and ineffective. Methods, usually
quantitative, were developed to describe an opera-
tional problem, to determine what information was
needed to solve it, and to analyze this information in
order to produce the best response. During World
War II, operations analysts in the armed services had
to use "existing" resources to solve tactical problems:
to employ radar countermeasures, to design effective
bombing patterns, to deploy destroyers around a con-
voy. From this experience, systems analysis grew to
select and evaluate "proposed" weapons systems. This
required that research (on what was technically feasi-
ble) match defense requirements (which sometimes
were not technically feasible).
  Systems analysis could, for example, assist in deter-
mining the need for an Army airlift or for a desired
mix of airlift and sealift to deliver a fighting force
overseas. The analysis would not include the design of
the transport aircraft; it would, however, consider the
performance requirements that the craft should
meet.7
  The same problems that had fostered the develop-
ment of systems analysis still existed in its application.
The most important of these were: (1) there was little
experience with the new weapons and delivery sys-
tems; (2) it was often difficult to experiment with
some of the real weapons and systems; (3) the techno-
logical complexity of new weapons grew rapidly; (4)
managing the development, procurement, and main-
tenance of new weapons systems became more diffi-
cult; and (5) there were severe time pressures to solve
problems or to counter immediate threats. The tech-
nique of systems analysis, however, dealt more effec-
tively with these problems than had prior planning
techniques. The hierarchical structure of the military
facilitated use of the method and enhanced its reputa-
tion as an effective technique of managing. Accept-
ance of the systems approach was also aided by the cir-
cumstance that it made public intervention in defense
decision making difficult, because it created the ap-
pearance that the public lacked technical expertise.8 A
Rand Corporation study, edited by E. S. Quade and
W. I. Boucher, provides a discussion of many of the
  
NUMBER 3

55



defense applications of systems analysis.9
  As military analysts gained experience with the new
techniques, they began to include national policy and
strategy in their analysis. Defense policy and strategy
were well suited to the systems approach, because the
policy goals were well defined in the 1940s and 1950s.
This higher level systems analysis (or policy analysis)
was later applied to determining the role of space sys-
tems in national defense and to comparing policies of
military superiority vs. military parity with the Soviet
Union. The mix of forces needed for national defense
has probably been the most intensive systems analysis
of policy.10
  Policy analysis has been used extensively by the Air
Force. One of the most influential studies was that
commissioned by the Air Force in 1951 to analyze the
most effective ways to acquire, construct, and use
strategic airbases in foreign countries. The Rand Cor-
poration's study determined that national security
would be higher and costs would be lower if the stra-
tegic airbases were located only in the United States.11
An Application
  During the early 1960s, California's population in-
creased so fast that 200,000 additional jobs were need-
ed per year. In addition, expected increases in the 20-
to 24-year-old age group would add another 59,000 to
65,000 persons per year to the labor force. Finally, in-
dustrial automation in California would also certainly
lead to increased productivity but not lead to a com-
mensurate increase in demand for labor. Even without
cutbacks in aerospace work, the already high unem-
ployment rate in California would increase if some ac-
tion were not taken by the state.12
  The increasing population and unemployment,
primarily concentrated in the urban areas, worsened
already substantial urban problems. Pollution and
poor mass transportation, inadequate housing, needs
for additional welfare assistance, and land use plan-
ning all were becoming more urgent. So also was ex-
ternal pressure on state and local governments to con-
struct housing, to subsidize mass transit, and to
establish training programs.
  Staff members of the California Department of
Finance wished to encourage economic development
that would provide revenue to the state, employ as
many of its citizens as possible, and be diversified and
healthy. They were aware that the current revenue
base was adequate but fragile.13 The economy, they
also knew, was too dependent on space and defense

contracts. At its peak in 1963, aerospace employment
was at 510,200. In 1964, that number had dropped by
26,300. Concurrent with the drop in space work, de-
fense contracts awarded to California firms decreased
by 13 percent in 1964. One analyst anticipated in 1966
that defense work would stabilize but predicted that
space employment would continue to decline.14
  The stimulus to the development of the "California
Studies" had come in December 1962, when the
planned $2 billion Skybolt Program,15 of which Cali-
fornia firms were to be substantial beneficiaries, was
cancelled. Governor Edmund G. Brown formed an
Advisory Panel on Aerospace and Electronics In-
dustries to suggest how to soften the impact of this and
future cutbacks. The panel was composed of some 14
executives of major defense companies and several
economists invited to study problems of increasing
unemployment and future defense reductions. In the
summer and fall of 1964 the Economic Development
Agency and the Department of Finance began to in-
vestigate with this panel how the state might ease the
impact upon the economy.16
  Even though there was agreement that the aero-
space industry and the state government could help
each other, the form of mutual assistance to be under-
taken was not obvious. Aerospace firms were accus-
tomed to responding to requests to solve specific
technical problems. Procurement and techniques of
project management similar to those used by the
Department of Defense and the National Aeronautics
and Space Administration would increase the prob-
ability of obtaining a useful product—but very few
knew what the "product" should be. As discussions
progressed, it became evident that it was a "process"
instead of a product that needed to be identified. The
problems confronting California were complex and
their potential solutions remained undefined. It was
thought that aerospace firms could use systems anal-
ysis to define the problems and to design programs
that might solve them.17
  A plan developed out of meetings between the panel
and the state agencies. The state was to offer requests
for proposals (RFP) to the aerospace industry to study
the problems of the state. In preliminary discussions
aerospace firms were receptive to the idea.
  Governor Brown scheduled a one-day symposium on
"California's Search for New Sources of Economic
Growth," at which he formally announced The Cali-
fornia Aerospace Study Program. The announcement
was made on 14 November 1964, with much fanfare.
A great deal of publicity accompanied the issue of the
  
56

SMITHSONIAN STUDIES IN AIR AND SPACE



RFPs. They were sent on 17 and 18 November 1964,
and proposals were due 18 December 1964. The stud-
ies were to have a fixed price of $100,000 each, theo-
retically to allow state administrators to evaluate all
proposals on performance alone.18
  RFPs were employed to guide the prospective con-
tractors, but state administrators were not sure what
they wanted from the bidders. They did not know the
dimensions of the problems, nor did they prescribe the
kinds of solutions that would be needed. The RFPs
were designed to encourage bidders to propose crea-
tive approaches and to deal systematically with the
problems. The wording of the requests was therefore
as general as possible.19 While this allowed maximum
flexibility, it gave the contractors few guidelines and
reduced the chances that their products would be of
use to the state.
  Many aerospace firms were unaccustomed to deal-
ing with vague objectives, but several were interested
in diversifying. Thirty-one bidders, substantially less
than half of which were California aerospace firms,
submitted 50 proposals in the four areas. Despite this
fact, ultimately, all four of the studies were awarded to
large aerospace firms based in California. All of the
companies that were awarded contracts had planned
continuous efforts to diversify their product lines and
had a high percentage of their employees engaged in
research and development. Contracts were let by Feb-
ruary of 1965, each for $100,000 and each for a dura-
tion of nine months.20 The projects were the follow-
ing:
  1. The Lockheed Missiles & Space Company under-
took to design a statewide information handling
system and to develop a plan to implement it.
  2. Space General Corporation (a subsidiary of
Aerojet-General Corporation) explored the feasibility
of applying techniques of systems engineering and of
operations analysis to social problems. They were also
to recommend a program for prevention and control
of crime.
  3. The Aerojet-General Corporation was to assess
the suitability of systems analysis and systems engineer-
ing to manage waste in California. They were also to
outline research and development to be undertaken in
the first stage of a broad program.
  4. North American Aviation, Inc., was to specify
systems to solve problems of transportation.21
  The studies were funded without a specific ap-
propriation from the legislature—a legal, but im-
politic action on the part of the governor. Money was
allocated from appropriated, but unspent funds. In

some cases, such as in the transportation study, the
original purpose of the appropriation could have been
construed broadly to include the study. In others,
however, especially those on waste management and
information systems, most of the funds were clearly in-
tended for other purposes. Legislative leaders James
R. Mills, Jesse Unruh, and others voiced some resent-
ment, which was not focused on the new uses for the
funds but on the way in which the legislature had been
ignored in negotiating the entire program.22 This
would not aid the fortunes of the recommendations in
the study.
  Despite these grumblings, the four contracts were
awarded, and the work was performed within the pre-
scribed time. The "California Statewide Information
Study" contained a general description of patterns of
flow and storage of information, a proposed statewide
information system and the benefits that would result
from implementing it. The study also contained a
step-by-step plan to implement the new system, the
associated costs, and suggestions of how to finance its
development and operation.23
  "Prevention and Control of Crime and Delinquen-
cy" described existing methods for dealing with crime
and delinquency and recommended a plan for devel-
oping a new system of criminal justice. A "Master
Plan" included schedules of eight programs, ranging
from more efficient administration of the existing
system to innovative programs such as one to create
more jobs. The schedule for implementation called for
a phase to study the need for each program and con-
tained estimated costs of the Master Plan.24
  Aerojet-General's "California Waste Management
Study" described the problems of refuse and sewage
handling and evaluated methods to dispose of it. The
study described the systems approach and its applica-
bility to "waste management," compared costs and
listed the functions of several systems, and produced a
plan to design one comprehensive system.25
  The "California Integrated Transportation Study"
was conducted under the joint sponsorship of the state
of California and the Bureau of Public Roads of the
U.S. Department of Commerce. Researchers devel-
oped a model and demonstrated its applicability to
several problems. They then planned a program for an
integrated transportation system in California.26
  A fifth contract had been awarded to System Devel-
opment Corporation (SDC). This company served as
liaison for the firms engaged in research and, perhaps
more importantly, sought federal funds to implement
the studies. After their submission, SDC was also re-
  
NUMBER 3

57'



tained to evaluate the final studies.27 SDC reviewed the
four final reports for soundness of technological and
methodological assumptions and for feasibility in ap-
plying systems analysis to major public problems and
reported the results to the state liaison officer.
  One report, North American's "Integrated Trans-
portation," was rated excellent; two reports—Aerojet-
General's "Waste Management" and Space General's
"Prevention and Control of Crime and Delinquen-
cy"—were ranked lower but deemed well within the
range of acceptability; and the Lockheed Missile &
Space Company's Information System received special
positive comment. This study, according to SDC, rep-
resented an enormous amount of work—far more than
the state could have expected for its $100,000 invest-
ment.28 (It is interesting to note that the most
favorably reviewed studies were those that addressed
the more technical problems.) In December of 1965,
the Department of Finance indicated its satisfaction
with the contents of the four studies:
. . . [they] in and of themselves, have been notably successful. They
have introduced new ways of thinking, new concepts to many of the
agencies involved. They have generated even in offices not involved
in the studies themselves, an excitement about these and other
things that government can do. The specific recommendations of
the studies have a good chance, by and large, of implementation.29
  Little was done subsequently, either in follow-up
study or in implementing program recommendations.
John Gilmore and others reviewed the California aero-
space experience in July 1967, analyzing the successes
and failures of the studies. The following section is
derived mostly from their work, supplemented by
more recent information.30
Analysis of Significance
  It is unclear whether the studies were beneficial to
the four contractors who performed them. Each spent
far more than $100,000 on the studies. (Estimates of
the out-of-pocket costs to each company vary between
$50,000 and $200,000). Ten years later, there have
been no profitable follow-on studies. Although aero-
space firms understood that it might take five years for
this kind of work to become profitable and that the
studies generated valuable experience, many underes-
timated the problems of developing the new research
market, especially in convincing the legislature to fund
additional work. Nonetheless Space General and
Aerojet-General received second contracts from the
state of California, although neither was a follow-on to

earlier work. All four original contractors and other
defense firms have expanded their work in systems
analysis to civilian problems, and most are performing
work for state, local, and federal governments. Fund-
ing levels remain lower than originally envisioned.
  The scarcity of follow-on work is due in part to the
character of the legislature, the nature of the prob-
lems, and the performance of the original contractors.
Criticisms of the studies most frequently encountered
were cited by Gilmore:
  1. The four reports sometimes showed poor knowl-
edge of the subject, as evidenced by incomplete, in-
adequate, or incorrect data and incomplete searches
of the literature.
  2. Some recommendations were politically naive or
impractical to implement; the phases of implementa-
tion or integration were often underemphasized.
  3. There was too much emphasis on engineering
and not enough attention paid to social and institu-
tional aspects.
  4. The four studies did not attempt to identify
problems and then to develop methods to solve them.
Rather, the contractors attempted to solve these com-
plex problems in the nine months allotted—an impos-
sible task. Consequently they failed at both tasks.
  5. The contractors provided few alternative solu-
tions. At most, the contractors talked about two or
three different programs, but these generally repre-
sented different levels of funding. Contractors typical-
ly devoted most of their analysis to one solution.
Regardless of their merits, these criticisms represented
the perceptions of many potential customers in state
and federal civil services.
  It should be noted that the studies earned praise,
especially for (1) their comprehensive scope in "ap-
proaching" the problems and (2) the freshness of their
recommendations, as compared to traditional efforts.
A good example of this creativity lay in the study deal-
ing with crime and delinquency. A series of overlays
provided a summary of socioeconomic variables that
showed high correlations with crime. The chart iden-
tified, in a graphic manner, the Watts area of Los
Angeles as a "hot spot" in advance of the riots of
August 1965.31
  From the state's standpoint, the four studies certain-
ly created local and national publicity—in fact, they
stirred international interest. While most of the pub-
licity was favorable, some criticisms from academics
and professionals increased with time, as study recom-
mendations were not implemented.
Ida R. Hoos, perhaps the best known and most ar-

58

SMITHSONIAN STUDIES IN AIR AND SPACE



ticulate critic of the "California Studies," has stated
that the techniques of systems analysis cannot success-
fully be applied to social problems. In an article pub-
lished in Datamation magazine, Hoos asserted:
      the very characteristics which distinguish social from other
species of systems render them resistant to treatment that tries to
force them into analytically tractable shape: 1. They defy definition
as to objective, philosophy, and scope. For example, what kind of
definition of a welfare system can be regarded as valid— that which
encompasses the shortcomings of other systems, such as health,
education, employment, or the one which focuses on individual in-
adequacy? A definition depends on the point of view and the ideo-
logical posture. The system looks very different to the administrator,
the recipient, the black power monger, the social critic, and the pol-
itician.
  2. "Solution" of social problems is never achieved. You do not
"solve" the problems of health or transportation. Consequently,
where you start and where you stop is purely arbitrary, usually a
reflection of the amount of money the government has to fund the
particular analysis.
  3. Despite the semblance of precision, there are no right or
wrong, true or false solutions. Consequently, it is presumptuous to
label as wrong anything being done now and right that which looks
good on paper. By concentrating on miniscule portions or isolated
variables simply because they are quantifiable, the technique may
actually lead to results which are irrelevant and inappropriate.
Assignment of social costs and social benefits is an arbitrary matter,
and even dollar cost/benefit comparison is a matter of interpreta-
tion.32
A similar though more extensive criticism of the stud-
ies appears in Hoos's book Systems Analysis in Public
Policy.33
  When viewed from the standpoint of technical
soundness and political acceptability, the results were
mixed. The wide scope of the problems tackled and
limits of time and money contributed to their failings.
  It is not possible to determine if the studies enabled
or hastened administrative reforms in the state govern-
ment. Prior to the 1966 California gubernatorial elec-
tions, there were indications that the studies helped

those reforms. With the election of Ronald Reagan,
Brown's goals (which had included the use of systems
analysis in state management) were no longer consid-
ered desirable.
  Albert Lipson, now an employee of the Rand Cor-
poration, was director of research for the California
Assembly when the "California Studies" were com-
pleted. Lipson states that the studies provided broad
new perspectives from which to analyze major prob-
lems.34 But the studies were not of lasting benefit to
California, in part because the contractors did not
provide the state what it needed. Contractors put the
bulk of their efforts into solving problems rather than
focusing on methods to solve problems. While their
solutions may have been excellent (hence the praise of
technical method and value received for dollars in-
vested) , the state was not prepared to implement
them.35
  Praise of technical content and criticism of political
inviability are not inconsistent. The state administra-
tion—like most state and local governments—did not
have an overall program to analyze policy. Such a pro-
gram would typically contain political and technical
components—both of which have legitimacy in the
process of state governance. Within such a framework,
the studies could have been valuable.36 But they had
been undertaken merely as an experiment, as a first
attempt in using a systematic approach to solve state
problems. As a result, there was no plan for what to do
with the studies once they were completed.37 Lipson
states that the governor commissioned a team almost
as an afterthought to review the studies and "figure
out what to do with them." It made little difference
whether Brown or Reagan was governor. The state of
California did not have then, nor does it have now, a
system of decisionmaking within which to fit such re-
search.38 The "California Studies" remain interesting
artifacts rather than useful documents.
  
Notes

  1. Elliot F. Beideman, State Sponsorship of the Application of
Aerospace Industry Systems Analysis for the Major Public Problems
of California (Ph.D. dissertation, School of Business Administra-
tion, University of Southern California, 1966), page 1.
2. Beideman, page 2.
3. Beideman, pages 3-4.
  4. E. S. Quade, Systems Analysis Techniques for Planning-
Programming-Budgeting (Santa Monica, California: The Rand
Corporation, 1966), page 28.
  5. John S. Gilmore, et al., Defense Systems Resources in the
Civil Sector: An Evolving Approach, An Uncertain Market (pre-

pared by University of Denver Research Institute for U.S. Arms
Control and Disarmament Agency, Washington, D.C: Govern-
ment Printing Office, July 1967), page 9.
6. Gilmore, pages 5-6.
  7. J. Gordon Milliken and Edward Morrison, Aerospace Man-
agement Techniques (Denver, Colorado: University of Denver
Research Institute, November 1971), page 48.
8. Gilmore, pages 13-16.
  9. E. S. Quade and W. I. Boucher, editors, Systems Analysis
and Policy Planning (New York: American Elsevier, 1968).
10.    Milliken and Morrisor, pages49-50.

NUMBER 3

59



  11. Bruce L. R. Smith, The RAND Corporation (Cambridge,
Massachusetts: Harvard University Press, 1966).
12. Beideman, pages32-34.
  13. Interview, Daniel Luevano, formerly chief deputy director,
Department of Finance, state of California, 29 July 1975.
14. Beideman, pages 25-26.
  15. Skybolt was an air-to-surface missile developed for the
Boeing B-52. The program was cancelled by the Department of
Defense because of cost overruns and an unsuccessful test program.
  16. Ronald P. Black and Charles W. Forman, Transferability of
Research and Development Skills in the Aerospace Industry (Falls
Church, Virginia: Analytic Services, Inc., 1965), foreword and
pages 1-9.
  17. Interview, William Lipman, formerly head planner, Office
of Planning, state of California, 30 July 1975.
18. Black and Foreman, Transferability.
19. Lipman interview.
20. Beideman, pages 129-131.
21. Gilmore, page 39.


22. Beideman, pages 56-61.
23. Beideman, pages 180-181.
24. Beideman, pages 222-223.
25. Beideman, pages288-289.
26. Beideman, pages 336-337.
27. Black and Forman.
28. Beideman, pages439-440.
29. Beideman, page 443.
30. Gilmore.
31. Beideman, page 266.
  32.	Ida R. Hoos, "A Realistic Look at the Systems Approach to
Social	Problems,"Datamation (February, 1969), page223.
  33. Ida R. Hoos, Systems Analysis in Public Policy, A Critique
(Berkeley, California: University of California Press, 1972).
  34. Interview,   Albert   Lipson,   The  Rand  Corporation,   San
Francisco, California, 13 November 1974.
35. Lipman interview.
36. Lipson interview.
37. Luevano interview.
38. Lipson interview.
33. 
-fr U.S.GOVERNMENT PRINTING OFFICE: 1979 0-290-858



REQUIREMENTS FOR SMITHSONIAN SERIES PUBLICATION
     Manuscripts intended for series publication receive substantive review within their
originating Smithsonian museums or offices and are submitted to the Smithsonian Institution
Press with approval of the appropriate museum authority on Form SI-36. Requests for
special treatment—use of color, foldouts, casebound covers, etc.—require, on the same
form, the added approval of designated committees or museum directors.
     Review of manuscripts and art by the Press for requirements of series format and style,
completeness and clarity of copy, and arrangement of all material, as outlined below, will
govern, within the judgment of the Press, acceptance or rejection of the manuscripts and art.
     Copy must be typewritten, double-spaced, on one side of standard white bond paper,
with iy4" margins, submitted as ribbon copy (not carbon or xerox), in loose sheets (not
stapled or bound), and accompanied by original art.  Minimum acceptable length is 30 pages.
     Front matter (preceding the text) should include: title page with only title and author
and no other information, abstract page with author/title/series/etc., following the establish-
ed format, table of contents with indents reflecting the heads and structure of the paper.
     First page of text should carry the title and author at the top of the page and an unnum-
bered footnote at the bottom consisting of author's name and professional mailing address.
     Center heads of whatever level should be typed with initial caps of major words, with
extra space above and below the head, but with no other preparation (such as all caps or
underline).    Run-in paragraph heads should use period/dashes or colons as necessary.
     Tabulations within text (lists of data, often in parallel columns) can be typed on the text
page where they occur, but they should not contain rules or formal, numbered table heads.
     Formal tables (numbered, with table heads, boxheads, stubs, rules) should be sub-
mitted as camera copy, but the author must contact the series section of the Press for edito-
rial attention and preparation assistance before final typing of this matter.
     Taxonomic keys in natural history papers should use the alined-couplet form in the
zoology and paleobiology series and the multi-level indent form in the botany series. If
cross-referencing is required between key and text, do not include page references within the
key, but number the keyed-out taxa with their corresponding heads in the text.
     Synonymy in the zoology and paleobiology series must use the short form (taxon,
author, year.page), with a full reference at the end of the paper under "Literature Cited."
For the botany series, the long form (taxon, author, abbreviated journal or book title, volume,
page, year, with no reference in the "Literature Cited") is optional.
     Footnotes, when few in number, whether annotative or bibliographic, should be typed
at the bottom of the text page on which the reference occurs. Extensive notes must appear at
the end of the text in a notes section. If bibliographic footnotes are required, use the short
form (author/brief title/page) with the full reference in the bibliography.
     Text-reference system (author/year/page within the text, with the full reference in a
"Literature Cited" at the end of the text) must be used in place of bibliographic footnotes in
all scientific series and is strongly recommended in the history and technology series:
"(Jones, 1910:122)" or ".  .  . Jones (1910:122)."
     Bibliography, depending upon use, is termed "References," "Selected References," or
"Literature Cited." Spell out book, journal, and article titles, using initial caps in all major
words. For capitalization of titles in foreign languages, follow the national practice of each
language. Underline (for italics) book and journal titles. Use the colon-parentheses system
for volume/number/page citations: "10(2):5-9." For alinement and arrangement of
elements, follow the format of the series for which the manuscript is intended.
     Legends for illustrations must not be attached to the art nor included within the text but
must be submitted at the end of the manuscript—with as many legends typed, double-
spaced, to a page as convenient.
     Illustrations must not be included within the manuscript but must be submitted sepa-
rately as original art (not copies). All illustrations (photographs, line drawings, maps, etc.)
can be intermixed throughout the printed text. They should be termed Figures and should
be numbered consecutively. If several "figures" are treated as components of a single larger
figure, they should be designated by lowercase italic letters (underlined in copy) on the illus-
tration, in the legend, and in text references: "Figure 9b_." If illustrations are intended to be
printed separately on coated stock following the text, they should be termed Plates and any
components should be lettered as in figures: "Plate 9JD." Keys to any symbols within an
illustration should appear on the art and not in the legend.
     A few points of style: (1) Do not use periods after such abbreviations as "mm, ft,
yds, USNM, NNE, AM, BC." (2) Use hyphens in spelled-out fractions: "two-thirds." (3)
Spell out numbers "one" through "nine" in expository text, but use numerals in all other
cases if possible. (4) Use the metric system of measurement, where possible, instead of
the English system.    (5) Use the decimal system, where possible, in place of fractions.
(6)	Use day/month/year sequence for dates: "9 April 1976."  (7) For months in tabular list-
ings or data sections, use three-letter abbreviations with no periods: "Jan, Mar, Jun," etc.
     Arrange and paginate sequentially EVERY sheet of manuscript—including ALL front
matter and ALL legends, etc., at the back of the text—in the following order: (1) title page,
(2) abstract, (3) table of contents, (4) foreword and/or preface, (5) text, (6) appendixes,
(7)	notes, (8) glossary, (9) bibliography, (10) index, (11) legends.