Documenting Postwar Science: the Challenge of Change

Organisational and technological change present enormous challenges to archivists and others who must learn how to document the new phenomena. To meet the challenge, JCAST (the American Joint Committee on Archives of Science and Technology), in its 1983 report, recommended documentation research, that is, research to resolve archival problems. I will illustrate its first use in a project aimed at learning how to document the postwar government contract laboratories. The main subject of my paper will be the dramatic rise of the multi-institutional collaboration and the archival implications of changes in organisational structures, communications patterns, and records of research.

Documentation Research

Until the mid-twentieth century the site for research in physics and allied sciences was overwhelmingly academia. The Center's initial documentation strategy focused on this setting; its appraisal guidelines emphasised the value of correspondence files, research notebooks, and other records that characterised the prewar period. In its efforts to document prewar - and, therefore, largely academic - physics, the AIP worked in a cooperative environment of archivists. In fact, during the 1950s and 60s, there was a boom in archival programs in academia; many were initiated, others were strengthened. There is no doubt that the AIP Center's early efforts were successful because of the strengths of academic archival programs.

As the AIP's cooperative ties with academic archivists grew, the feeling of comfort that we were doing a good job of documenting postwar physics dimmed. More and more often we were reminded - by the press, meetings of the American Physical Society, and other reports - that some of the most distinguished work was taking place at nonacademic settings. The shifts were partly to corporate research laboratories, but most spectacularly, to the postwar phenomenon - government contract laboratories. There was little knowledge of what went on in nonacademic settings. Site visits to laboratories of all sorts made it clear that this postwar, nonacademic community was so dramatically different that it was impossible to state with any confidence how it ought to be documented. We found the following:

1. There were NO ARCHIVISTS at any of the labs.

2. There were NO HISTORIANS using postwar records. Actually, there were two or three - not enough to have an impact on saving the records. We felt certain the historians would be coming - looking for records - before too long, simply because these were major institutions in their own right and because a good deal of the research was of high significance.

3. RECORDS MANAGERS, on the other hand, were everywhere.

4. Those in the CORPORATE labs were likely to be PROFESSIONALLY trained. Their time was DEDICATED to records management.

   Those in GO-COs (government contract labs) were typically not trained as PROFESSIONALS. Their responsibilities were NOT CONFINED to records management; they might, for example, also be responsible for running the lab's cafeteria.

5. Every lab seemed to have RECORDS disposition SCHEDULES that determined how long records of a particular type could be retained. We wondered if these schedules were ANY GOOD.

6. There were MOUNTAINS OF UNSCHEDULED RECORDS and many reports of DESTROYED RECORDS.

7. NARA (the National Archives and Records Administration) took the position that an institution could be adequately documented merely by saving records at the top of the hierarchy. This so-called PYRAMID VIEW OF APPRAISAL had been based on the National Archives' experience with nonscientific agencies such as the State Department. We were confident it should not be applied to scientific labs, where much of the history is made by scientists farther down in the hierarchy.

8. We found TRADITIONAL NOTEBOOKS, BUT we ALSO found ELECTRONIC RESEARCH DATA. We had a lot of worries about the electronic data, especially about our ability to appraise its long-term usefulness and how it should be stored.

What could be done? What approach would be appropriate under the circumstances? We felt quite certain that traditional archival practices - for example, saving the papers of eminent individuals - would be inadequate. Because of the bulk of postwar records and new forms of evidence, continuing the practice of appraisal on a collection-by-collection basis would be ineffectual and costly. There was a realistic sense of urgency: historians would be fortunate if they arrived to use the records before records managers had scheduled them for destruction. What was called for was a new kind of research. JCAST would later call it documentation research.

What is meant by the term is research aimed at resolving archival problems. Documentation research involves two important stages. The first stage consists of systematic planning and research, and close interaction with records creators - physicists and science administrators in our case. The goals are several: first, to identify the most important organisational structures, functions, records creators, and events; next, to understand how and why records are created and used; and, finally, to identify the likely locations of valuable records. This research replaces the practice of appraisal on a collection-by-collection basis. There are a number of terms that could be applied to this stage. It has been called 'functional analysis', 'historical-sociological research', and - in some cases - the 'study of organisational sociology in real time'. Some Europeans and Canadians are calling it 'macroappraisal'. I use these terms interchangeably, but as an archivist my favourite term is 'macroappraisal'. The term clearly suggests that what we are talking about is the research and systematic analysis that must be carried out before doing the traditional, microappraisal of records. When we do that appraisal, it will not be the whole universe, but a carefully identified and potentially valuable subset of the records.

Phase Two of documentation research is devoted to policy and programmatic issues. This is the time to ask: What documentation is at risk under current procedures? This is the time for action to initiate new policies and procedures to secure valuable documentation. In documentation research projects of the AIP, this stage is critical. The AIP conducts the first stage to learn how to document an area. With that knowledge in hand, it issues formal policy recommendations to institutions that have control over the records.

The first documentation research project, completed a decade ago by the AIP Center, was a study of four Department of Energy (DOE) National Laboratories: Argonne, Brookhaven, Lawrence Berkeley, and Oak Ridge. The program was known as the "DOE Study." Rather than go into details, I will describe briefly how we picked our targets and designed the macroappraisal stage of our project and make a few observations on the effectiveness of these tactics and our project recommendations. To return to the problems I listed earlier:

1. NO ARCHIVISTS
2. NO HISTORIANS
3. RECORDS MANAGERS EVERYWHERE
4. GO-COs: NOT PROFESSIONALS, NOT DEDICATED
5. RECORDS SCHEDULES - ANY GOOD?
6. MOUNTAINS OF UNSCHEDULED RECORDS AND DESTRUCTION STORIES
7. NARA'S 'PYRAMID' VIEW OF APPRAISAL
8. TRADITIONAL NOTEBOOKS, BUT ALSO ELECTRONIC RESEARCH DATA

First, as you see, the corporate laboratories are off the list. Since each corporation is a separate entity with its own problems and practices, our research and recommendations would have to be individualised; that was a task beyond our resources. We could use these resources far more effectively with documentation research aimed at resolving problems surrounding Federal scientific records; this could improve the programs and policies of the National Archives. The most significant government contract laboratories creating Federal records were, and still are, the national laboratories of the DOE. Thus, by focusing on four DOE national labs and the National Archives, the ripple effect of our findings could improve records-keeping practices at other government scientific agencies.

The project historian and project archivist spent over eight staff-years working on site at the four laboratories. We did not solve the problems on our list by reviewing mountains of records [6]. During our first, macroappraisal stage, we employed extensive historical research and involvement with laboratory scientists. We prepared chronologies of the main events and programs (scientific and otherwise). We talked to the key players -scientists and administrators - about these events and programs and about the records created in the process. With their help, we located the best documentation. Through these and other tasks, project staff achieved an understanding of the ways government contract laboratories operate and of the information content of the records they generate.

Through its field work, the AIP study proved that valuable records had, indeed, been created, but could no longer be located and were, presumably, destroyed. (The destruction stories were true [6]). We also proved for the first time that valuable documentation of science and technology would not be captured by only saving the records at the top of the institutional hierarchy. (The National Archives' pyramid view of appraisal did not apply here [7]). In addition, we developed solid appraisal guidelines - one for policy and administrative files and a second for scientific files; by comparing the records that should be saved against the records schedule authorised by the National Archives, we showed that those schedules needed important revisions [5]. These and other findings gave credence to the project's policy recommendations addressed to the DOE labs, the DOE headquarters, and the National Archives.

Our single, most important recommendation was that professional archivists were needed at each national laboratory. Each of the four DOE laboratories was unique in some way. Although guidelines for records appraisal and reports on general patterns of administration and research are helpful, archivists were needed at each laboratory. The archivist's knowledge of the organisational structure and functions of his or her own institution must be the basis for decisions as to what documentation should be saved. I'm happy to say that a number of important DOE labs now have professional archivists [1] who typically supervise improved records management programs [3]. It is also true that the National Archives has become far more concerned about science and technology. They are currently upgrading their guidelines, prompted in large part by the DOE study. Best of all, there are now dozens of historians writing about postwar science [2].

Problems remain. Many important labs still lack archival programs and the National Archives is stretched thin in its ability to provide budget, staff, and space to secure valuable records. You'll note, too, that the item on electronic research data still remains on the problem list; we only scratched the surface in this area [8]. Nevertheless, there is reason for optimism: we learned how to document postwar government contract laboratories. In addition, we found that the methodology applied to other postwar laboratories, including corporate laboratories.

Organisational Change: the Rise of the Multi-institutional Collaboration

Spencer Weart (the AIP Center's director) and I decided to explore the extent to which institutions were joining forces to carry out significant research. To obtain an overview of the impact of large team multi-institutional research and the likelihood of the documentation being saved, I analysed 68 major events in physics and astronomy reported in the magazine Physics Today over the period from 1968 to 1973. The results of this preliminary study were dramatic. (1) Only one-sixth of the events were the work of a single physicist while over one-half involved four or more; (2) close to one-fourth of the events involved off-site research, primarily at accelerator facilities or astronomical observatories; (3) about one-third of the events involved two or more employers, and nearly half of these involved four or more employers; and (4) while academic employers were involved in over half of the events, the traditional setting of a single academic employer only accounted for one-fifth of all the events. It was clear that postwar science poses a far more complex web of institutional efforts to document than ever before. And, I need to point out that if - instead of 1968 to 1973 - I had examined the period ten years later or 20 years later, the findings would have been still more dramatic.

Since World War II, the multi-institutional collaboration has increasingly been the organisational framework for scientific research. Its impact has been most visible in the field of high-energy physics. In this discipline, members of a single collaboration may currently number in the hundreds, come from a score of institutions based in several countries, and take a decade to prepare and conduct an experiment at a unique accelerator facility. The experimental results may be of sufficient interest to excite the popular press and inspire the award of a Nobel Prize, so it is quite clear that future historians and other scholars will want to study the work. Equally significant multi-institutional research is conducted by teams in other fields of physics, in other fields of science, in major engineering projects, and in a growing number of other areas of modern society. There can be no doubt of the high importance of understanding how such work is, in fact, carried out.

Yet the process of large-scale collaborative research has been little studied by historians, sociologists, and other scholars. Nor have scientists, research administrators, and the public made any systematic analysis of this centrally important activity. Despite the significance of these transitory 'mini-institutions,' we were not able to find adequate records of any collaboration in a repository.

A documentation research project was clearly called for. In order to address the problems of documenting these transient institutions, we knew we had to first understand the process of collaborative research and how the records are generated and used. We decided to make a broad survey, the first of its kind, of the functioning of research collaborations involving three or more institutions. Because of the complexities, we thought the study should cover a number of fields where collaborative research played an important role. The choice of high-energy physics for Phase I was easy: it is the field in which multi-institutional collaborations have grown most dramatically and, in addition, the DOE study had given us familiarity with the functions of accelerator laboratories. We decided that, for Phase II, the field disciplines of space science, geophysics, and oceanography would provide a useful contrast to the laboratory discipline of high-energy physics.

In order to design the research and pick our targets, further investigation was needed. The main problem, of course, was how to obtain solid knowledge of the functions of these collaborations without spending months on-site. There were too many institutions involved in any one experiment and hundreds of experiments per year at accelerator laboratories! We clearly needed sophisticated sociological expertise to help design the macroappraisal stage of our research.

With the help of that advice, we decided to focus on key sites: for high-energy physics these would be accelerator laboratories, for space science and geophysics the sites would include space vehicles and oceanographic ships. For each site, we would select specific experiments representing main areas of the fields under study. Our period of study for all disciplines would be from the mid-1970s to the late-1980s. To guide and critique our work, we would assemble a working group of archivists, historians, and sociologists and, most important of all, distinguished scientists and science administrators from the discipline under study. Because the period under study is almost current, and collaborations have changed rapidly during the past two decades, I characterise this macroappraisal work as a historical-sociological study of organisational trends and their archival implications. It comes close to being a study of organisational sociology in real time.

The two-year study of high-energy physics research examined experiments carried out at five of the world's major accelerator laboratories. A broad picture of changes in the structure of collaborations - such as size and length of experiments - was obtained by using databases on high-energy physics experiments and publications. At a more detailed level, the project conducted interviews on 23 selected experimental collaborations. In selecting experiments, our aim was to cover a range of historical, sociological, and scientific parameters and a variety of archival situations. For each collaboration, we interviewed individuals at various levels to give us a full perspective on the project. We used a structured interview question set designed for high-energy physics. Nearly 300 interviews were systematically analysed for historical, sociological, and archival content.

The long-term study began in 1989. My observations will focus on organisational trends in high-energy physics, the subject of the first phase of our study - now completed. In high-energy physics, collaborations form around an experiment. Members build a detector for use at an accelerator, gather and analyse data, and publish findings. The collaborations consist of groups from universities and, often, from the accelerator laboratory itself. The individual groups are assigned specific responsibilities for building components of the detector and other tasks. The size and complexity of collaborations and the research facilities have grown dramatically. In 1992 the Superconducting Super Collider Laboratory received a proposal signed by 991 researchers for an experiment to build and use a detector; if approved, more institutions may join and more people will be added - especially postdoctoral and graduate students. Many large experiments take a decade or more from start to finish.

Collaborative research of this kind is Big - sometimes called Very Big - Science. More than anything else, it is the increased complexity and rising costs of apparatus for experiments that have contributed to the size of collaborations in high-energy physics. Any group with the ambition to build an expensive detector has had to convince physicists from other institutions or countries to join in the experiment.

The AIP study covered organisational structures and operational functions through all stages of an experiment. I would like now to focus on those structures and functions of collaborations in high-energy physics that are of particular interest because of their serious archival implications. These are data-gathering, communication, and the roles of the administrative leader known as spokesperson, and of the accelerator laboratory.

During our period of study, virtually all of the experimental research data were gathered in electronic format on magnetic tape. We found that these data are not needed for scientific purposes after a brief period of analysis; in addition, they are not useful for historical or other scholarly purposes. Archivists and records managers have reason to be concerned about retaining electronic data for some other disciplines, but they can confidently destroy them for high-energy physics. A small sample from each decade, preserved for exhibit purposes in a national laboratory or museum, would meet future needs.

In terms of organisational strategies and communication, we found that high-energy physics experiments require that all collaborations combine three organisational strategies. Each collaboration blends these strategies in ways that offer the best chance of handling its toughest difficulties. First, the laboratory is treated as an organisational headquarters to which the outlying institutions pass and receive information. This strategy becomes more dominant when the integration of detector components is so complex that the collaborating institutions send postdocs to the laboratory. Second, the collaboration determines the extent to which labour should be divided - with collaborators working independently - or duplicated. The need to duplicate efforts takes precedence when puzzling or controversial findings are claimed; here reproducibility of results from data analysis is essential. Third, collaborations take into account that individual researchers will likely make use of equipment they did not build and software they did not write; collaborations, therefore, set up an information pool that enables collaborators to take full advantage of what others have developed.

In any one of these strategies, the individuals or groups that need to communicate are likely to be different. Viewed over time, our study brings out one major trend: the intra-collaboration information network (for example, collaboration-wide mailings and technical memoranda) - this network has become increasingly formal and increasingly electronic. High-energy physicists now generally communicate through electronic media. The Japanese physicists I interviewed took credit for introducing the fax, and we found evidence of e-mail as early as 1982. The availability of electronic media certainly has enabled these large, far-flung mini-institutions to carry out their organisational strategies effectively and rapidly.

Another structural feature of high-energy physics experiments is the spokesperson; this person has been both an intellectual leader and an administrative convenience - an individual designated to speak for the collaboration to the laboratory and to inform collaborators of laboratory requirements. More recent collaborations tend to create administrative substructures to handle collaboration business, and they may change spokespersons over the course of their experimental runs. Managerial burdens of spokespersons have come to outweigh the opportunity for exercising scientific leadership and judgement.

Collaborations have traditionally designed and built components of their detectors largely at their home institutions and without oversight from the laboratories. More recently, however, a number of factors have contributed to shifting power and accountability from the university groups to the national laboratories. A decline in sophisticated laboratory and shop facilities at many American colleges and universities has led to the fabrication of more detector components at the accelerator sites. Beginning in the late 1970s, the laboratories have had tighter control over experiments - at least the larger, more expensive ones. In the United States, funding for building these experiments is more and more likely to come directly to the laboratories for distribution to the collaboration groups. Finally, most laboratories now require detailed contracts covering the responsibilities - of both the laboratory, and each of the institutional members of collaborations - for the performance of experiments. These findings provide strong indications of shifts of power and accountability within the organisational structure of collaborations.

We found that some particular circumstances affected (in a positive way) the creation or retention of valuable documentation. These include the size and geographical dispersal of institutions, the emergence of fax and electronic mail, the need to communicate with engineers, and the importance or controversial nature of experimental results. We are particularly delighted with the rise of e-mail; before e-mail, physicists tended to use the telephone, a practice that normally leaves no paper trail. E-mail is easily stored and thus often saved by physicists. Of course, it is also easily destroyed; early archival or records management intervention is essential.

The historical-sociological analysis of organisational trends is extremely useful for archival purposes, when combined with patterns of records creation, retention and destruction, and likely locations of records. In addition, our findings - coupled with records appraisal - show the main locations of valuable records. The most important are at the laboratories and in the hands of spokespersons; other records, which may be needed to document significant experiments, are with leaders of the individual institutional groups that are members of a collaboration. We established that a core set of records could provide adequate documentation of most collaborations. Among these are files of funding proposals; contractual agreements; and progress and other reports. Many of these have been traditionally kept by the labs, but have yet to be scheduled for permanent retention.

Our main concern is to secure the additional documentation needed for especially significant experiments. Take, for example, the importance of saving one full set of those collaboration-wide mailings I mentioned. Where should this documentation be located? The issue rests on the understanding of ownership and primary responsibility. Ownership is particularly rigorous when the records are Federal. But academic archives may question why they should save collaboration-wide records when their faculty was only one of a number of institutional groups on an experiment. If we are to document significant multi-institutional collaborations without undue duplication of effort, the community will need to develop a broader sense of responsibility and cooperation. This is a serious challenge.

In our appraisal guidelines, we point out that valuable records are increasingly created in electronic format. In addition to collaboration-wide mailings and e-mail, these include notebooks and correspondence of individual members, and logbooks of detector operations. Other archivists are dealing with the problems associated with preserving and 'migrating' records in electronic format. Their success would provide a major breakthrough in documenting modern science.

Whether valuable evidence is on paper or in electronic format, our findings show that archival action should be swift. Like other groups, most physicists keep documents only if they think they will be useful to themselves. Good records-keeping may be acknowledged by all as necessary while the experimental process is alive, but when the experiment is over, records can easily be neglected, forgotten, or destroyed. A decade from now, many of the records located by the AIP project may well be gone. To be most effective in documenting multi-institutional collaborations, future archival efforts should take place during the brief period of years when the records-keeping needs of the scientific collaboration coincide with the goals of archivists.

The product of the second, policy stage of the AIP study of high-energy physics is recommendations for action needed to secure valuable documentation. These are addressed to the laboratories, universities, and Federal agencies. Our single, most important recommendation is addressed to the laboratories. During our study we identified an opportune point of leverage that could have a major impact on documenting future experiments; this is the laboratory's contractual agreement with the collaboration. We ask that, once an experiment has been approved, the laboratory should have the spokesperson identify one of the collaboration members who would be responsible for its collaboration-wide records. In addition, where historical significance warrants, individuals should be named to be responsible for group-level documentation of innovative components or techniques. This information should be incorporated into the contractual agreement. Use of this simple mechanism would assist archivists everywhere by assuring that records would be available for appraisal and by providing information on their location.

I have discussed patterns of collaborations in the experimental, laboratory-based field of high-energy physics. The final reports are available from the AIP upon request.

For our current work in the disciplines of space science, geophysics, and oceanography, the methodology used in high-energy physics is being followed, with only a few changes. We are again conducting interviews on selected projects, preparing historical, sociological, and archival reports that will profile these fields, and investigating the possibility of preparing summary statistics. We have again formed a working group of distinguished scientists, science administrators, archivists, historians, and sociologists to help us to plan, and then to criticise our work.

I'd like to mention just one geophysics project we are studying - Leg 133 of the Ocean Drilling Program that drilled in the Great Barrier Reef. The leg was proposed by Australian scientists at the Bureau of Mineral Resources. This preliminary report, issued in 1990, shows that there were 29 scientists onboard the JOIDES Resolution, from 20 institutions in 10 countries. This number does not include the land-based scientists who participated, or any of the engineers, technicians, or other support staff.

Although we still have over a year of work to complete Phase II, we have already found a variety of patterns different from those in high-energy physics. One striking contrast is that the formation of collaborations in space science, geophysics, and oceanography is a longer, more political process; in addition to the Federal agencies themselves, National Academy of Sciences committees, Congress, and offices in the White House are likely to participate in the formative, pre-funding period; they are often joined by equivalent government agencies abroad. Another contrast is that in high-energy physics, scientists observe events they themselves created in the laboratory; in the field sciences of space science and geophysics, on the other hand, scientists observe the events or phenomena of nature. Again, in sharp contrast to high-energy physics, these disciplines do need data collected at different times and places for long-term scientific work. This means that the data are often of permanent scientific value and must be retained. Finally, a wider range of institutions appears necessary for a successful collaboration in space science and geophysics. Institutional collaborators run the gamut from government, government contract, and corporate to academic settings. In particular, there is far more use of contracts to industry. All of these differences have archival implications. However, we hope to find points of leverage - similar to those we found in high-energy physics - where our findings could have a major impact to improve documentation.

During the third and final phase of the study, we are proposing a comparative approach to extend and generalise our findings to other areas of science and technology - areas such as nuclear fusion, medical physics, microchip consortia, and clinical medicine. Building on the understanding gained from the earlier, tightly focused studies, we will examine and compare collaboration patterns - and the archival implications of those patterns - in each area studied. With the help of a specialist in organisational structures, we will develop, for each area, a 'typology', that is, a classification scheme for the functions and operations of collaborations. We will use these typologies as the basis for an overall typological study of collaborative research in science and technology. We will then draw general conclusions.

With this knowledge in hand, we will begin the most important policy stage of the entire documentation research project. We will develop policies and procedures that should help to secure important documentation of multi-institutional collaborations in any field of science and technology. Our recommendations will be addressed to a broad range of institutions and to policy-makers in the scientific and archival communities.

Conclusions

I hope that I have given some idea of how valuable systematic surveys of organisational structures and operational functions can be, for finding ways to resolve archival problems. As I have illustrated, documentation research can be usefully applied to a study of just one institution or to a structure involving many institutions. This kind of research is exciting and effective in terms of costs and improved documentation for future scholars.

Published by ASAP on ASAPWeb, 2 August 1997.
Designed by Tim Sherratt

Updated by: Elissa Tenkate
Date modified: 25 February 1998