A BILL TO PROVIDE A STANDARD REFERENCE

DATA SYSTEM

WEDNESDAY, JUNE 29, 1966

HOUSE OF REPRESENTATIVES,

COMMITTEE ON SCIENCE AND ASTRONAUTICS,

SUBCOMMITTEE ON SCIENCE, RESEARCH, AND DEVELOPMENT,

Washington, D.C.

The subcommittee met, pursuant to adjournment, at 10:25 a.m., in room 2325, Rayburn House Office Building, Hon. Emilio Q. Daddario (chairman of the subcommittee) presiding.

Mr. DADDARIO. This meeting will come to order. Our witnesses this morning will be Dr. J. Herbert Hollomon, Assistant Secretary of Commerce for Science and Technology, and Dr. Allen V. Astin, Director of the National Bureau of Standards. I think in order to proceed more expeditiously we will have Dr. Hollomon give his statement, to be immediately followed by Dr. Astin. We will hold our questioning until both statements have been given unless there is something which any of the members would like to have clarified as we go along. If you would proceed, Dr. Hollomon. We are happy to have you both here.

STATEMENT OF DR. J. HERBERT HOLLOMON, ASSISTANT SECRETARY OF COMMERCE FOR SCIENCE AND TECHNOLOGY

Dr. HOLLOMON. That is perfectly all right. We are always happy to be here. We have a warm feeling in our hearts for this committee, both in terms of its actions and the fact that you are so interested in the kind of things that we are interested in.

Mr. Chairman and members of this subcommittee, a serious inefficiency in our methods of application of science and technology to meet national needs is the lack of an adequate system for making expertly evaluated data on the properties of substances readily available to the nation's scientists and engineers. The process seems simple; data must be extracted from the world's literature, their reliability evaluated, and then they must be put in the hands of the man or woman who is going to use them. In practice the difficulties are great, as I shall describe later.

Fortunately, something can be done, and is being done, to reduce this inefficiency. We are here today to urge favorable action on legislation to further the efforts of the Department of Commerce, through the National Bureau of Standards, to serve this urgently felt need of the Nation.

The importance of critically evaluated data on the physical and chemical properties of substances and their interactions-commonly called standard reference data-is fully recognized by the technical community but is not well understood or appreciated by those who have not spent a great deal of time making scientific and engineering calculations. Therefore, I would like first to give a brief explanation of why standard reference data are important and how they are produced. Then I should like to describe the current status of national efforts to produce and disseminate standard reference data and explain why we are seeking this legislation to expedite the national program. Scientists and engineers all over the world measure the properties of substances and their behavior when interacting with each other and with energy in its various forms. The substance may be anything from a subnuclear fundamental particle, to a nucleus, an atom, a molecule, or a complicated mixture or solution.

The results of the measurements are numbers whose values depend upon the standards maintained in this country by the National Bureau of Standards. These numbers are published in scientific journals, reports, handbooks, and other publications. Therefore, the numbers are available to anyone who can locate them. But it is often extremely difficult to locate a specific number in the literally millions of pages of scientific literature, and, once located, also difficult to determine just how reliable the number is.

The problem is complicated by the fact that often more than one researcher works in the same field (and at different times) each coming up with his own measured value for the same property. Only a specialist in the field can tell which is most likely to be correct. The problem, therefore, is to

(1) Extract the necessary data from the literature;

(2) Determine the accuracy and reliability of the data through a process of critical evaluation; and

(3) Make the evaluated data readily available to users.

Hence, there is a need for standard reference data-called reference because scientists and engineers repeatedly refer to them in their work, and called standard because differing values are critically evaluated by the most competent scientists in the field who then select and certify a single value or range of values as the best or standard one. The critically evaluated data may then be used with maximum confidence, though they may always be revised in light of new knowledge.

Of the approximately million and a half scientists and engineers in the United States, about a quarter of a million are electrical engineers. How this group, for example, uses data illustrates the need for standard reference data. These engineers must use such properties as electrical resistivity, thermal conductivity, magnetic permeability, and melting point in their everyday work.

These and other properties enable electrical engineers to design communications devices, such as radios, TV, microwave systems, telephone systems, or electrical components for use in automobiles, in the home, in the factory, or in highly sophisticated space vehicles, where the components have to work reliably under extreme conditions.

These engineers also design power transmission systems, in which they must include safety controls-devices that cut off power when an overload or other dangerous possibility occurs.

Hundreds of thousands of engineers are concerned with transportation-motor vehicles, railroad and subways, airplanes, ships. In their daily work they need to know properties of alloys, metals, rubber, plastics, fuels, and a host of other materials.

In dealing with other great national problems, such as pollution, corrosion, safety, health, or contamination, engineers and scientists depend upon reliable and readily available values for the properties of materials to do their job well. For example, what are the properties of fuels which bear on smog control? What are the properties of detergents, especially how they break down in chemical reactions that render them harmless?

The melting point of a given ceramic material more specifically illustrates the need for reliable data. A team of engineers has the assignment of designing a space vehicle that will successfully withstand the exceedingly high temperature generated by friction as the vehicle reenters the atmosphere from a trip in space.

The design of this vehicle must take into account many considerations, among which are: (1) How much heat will friction generate on a given material with specific surface and other characteristics at a given speed in a given atmospheric density; (2) how much heat will the material tolerate before it breaks down or melts; (3) how can this heat be contained, dissipated, or insulated to protect the occupants or instruments in the vehicle.

If the necessary data to make these calculations are uncertain or unreliable, or hard to come by within the time available, the engineers are faced with undesirable alternatives:

They can make a new measurement to determine the needed number, thus duplicating work that has already been done, and with no guarantee of greater accuracy, or

They can overdesign the vehicle to assure safety. Thus, if the unevaluated data available shows a range of melting points for the particular ceramic material from 1500° to 2000° C., they would use the lowest one to be safe. But if they knew that a higher melting point were a reliable number, they could design accordingly.

Uncertainty and inaccessibility, therefore, are costly; they waste money, time, and scarce professional manpower; they cause delays in and sometimes abandonment of projects.

The present lack of a comprehensive, effective standard reference data system costs the Nation hundreds of millions of dollars each year. This is so because the work that an integrated, comprehensive, Standard Reference Data System could do is already being done. Nearly every member of the technical community does part of the job himself-piecemeal, uncoordinated, and usually less effectively than if done by an expert. Properly operated at full potential, a Standard Reference Data System could return in our estimate $20 to $200 for each dollar spent on it.

The products of the Standard Reference Data System are valuable to the technical community because numerical data are made both more readily accessible and more reliable. Dollar benefits result from (1) savings of time of users in searching through scientific reports for numerical data; (2) savings of time in evaluating and selecting most reliable answers from among those found in scientific report; (3)

savings of time and materials spent in unnecessary measuring of properties of substances for which the data actually reported in the literature could not readily be found; (4) savings of time, equipment, and materials through use of better-more reliable-data.

Some of these savings-those in the first two categories can be estimated quantitatively. An illustration is the example of NBS Circular 500, "Selected Values of Chemical Thermodynamic Properties," issued originally in 1952 and now being revised. This volume contained several thousand values of certain fundamental thermodynamic properties of all the elements, all inorganic compounds for which data were available, and organic compounds containing one or two carbon atoms. Over 7,000 copies of this book were sold-to scientists and engineers in academic laboratories, industrial laboratories, and government laboratories. Using conservative assumptions about the number of persons using each volume and the amount of time saved at each use because the individual did not have to search and evaluate for himself, we estimate that the equivalent value of this one volume to the economy of the United States has been $50 million. The cost of producing this publication was about $250,000. The ratio of benefits to cost in this case is 200 to 1-an unusually favorable ratio.

It is this type of estimate, which can be made for numerous similar works now in existence, which leads us to the conclusion that each dollar spent on producing standard reference data will save the economy $20 to $200. I must point out that these savings will be difficult to find. They will be made up of thousands of small timesavings, equivalent to a few dollars here and a few dollars there. No one will file a cost-saving report with his administration listing an item of $30 for 3 hours that he didn't spend in the library because a compilation of evaluated data was available to him.

The estimates just described do not take into account the value to the economy of the availability of better data. These values are inestimable. How would one determine, for example, how many manufacturing plants had to be designed with broader tolerances because the available data were less reliable than they might have been? How does one determine how many missile shots have failed because incorrectly evaluated data were used in the design of some component? Although such incidents cannot be definitely identified, every scientist and engineer is confident that they occur and that they are costly.

Important as such benefits are. they are perhaps matched in importance by the guidance provided to measurement practices by the data compilations. The thorough critical evaluation of sources of uncertainty in measurement technique inevitably leads to an upgrading of the quality of subsequent measurements in laboratories all over the country.

This effect has already begun to be felt as a result of the present NBS program. Further, by pointing out gaps in the availability of data and by identifying key properties for which higher precision is required, data evaluations serve as a means for an experimentalist to determine which measurements deserve high priority in his program. The question may well be asked: "If these compilations are so important, why is an enhanced program needed to make sure that they are available?" The answer to this question is the observed fact that

existing mechanisms for producing critically evaluated compilations have not been able to keep up with the flood of new data appearing in the literature.

Throughout the history of recorded science, numerous data compilations have been produced, largely in response to an urgently felt need of some part of the technical community. Some compilations were one-shot projects, resulting in a product that was never updated; others have been continuing activities lasting over many years. Some were sponsored by private organizations. There was no coordination or standardization of format or quality, and in some technical areas there was extensive duplication.

In recent years, murmurs of concern over the situation rose to a clamor that something be done. Committees of the President's Science Advisory Committee and of national professional societies carried out studies and made recommendations. The National Academy of Sciences attempted to stimulate new and expanded activities.

Mission-oriented agencies initiated crash projects to satisfy longstanding needs of which they became newly aware. The Coordinating Committee for Material Research and Development of the Federal Council for Science and Technology urgently recommended that a comprehensive program be started. The National Bureau of Standards laid plans for increasing the level of effort in its organization. Recognition of the problem and the cost to the Nation of failing to solve it are the reasons behind the recommendation that a Standard Reference Data System be established. This system came into being in 1963 through action of the Federal Council for Science and Technology and the President's Office of Science and Technology.

Responsibility for its administration was assigned to NBS. The Standard Reference Data System is considered to be one of the components of a broad national scientific and technical information system now being developed by the Committee on Scientific and Technical Information (COSATI) of the Federal Council for Science and Technology.

Briefly, the responsibilities of NBS are the following: (1) Promote compilation of evaluated data; (2) coordinate related work under the auspices of all Government agencies; (3) establish standards of quality for all products of the system; (4) operate a Standard Reference Data Center at NBS; and (5) establish standards of methodology and such other functions as are required to insure the compatibility of all units of the Standard Reference Data System. Dr. Astin's statement provides a description of the steps he has taken to discharge his responsibilities for the program.

Because of the universally recognized value of data compilations, scientists in other countries have also undertaken systematic programs. We have already had productive discussions with responsible officials in the United Kingdom, France, Germany, the U.S.S.R., and Japan regarding possible cooperative or joint programs.

The International Council of Scientific Unions has recently established a committee to serve as a coordinating body and channel of communications. The Department of Commerce intends in the subsequent development of the standard reference data program to maintain close and cooperative relationships with effective groups throughout the world, within appropriate foreign policy guidelines.

65-891-66- -3