Relation of the Federal Government to Research 



search, in response to a request, addressed to aj^encies 

 carrying on research in the natural sciences, for com- 

 ments on problems connected with their research work. 

 If data for an agency were compiled from published 

 or indirect sources, these data were formally trans- 

 mitted to the head of the agency for confirmation. 

 The amounts in this study given for total research 

 expenditures of each agency are therefore figures that 

 have been confirmed by the agencies themselves, ex- 

 cept where otherwise indicated or where an agency has 

 been credited with research paid for, but actually car- 

 ried on by the one agency which specifically requested 

 that such transferred amounts be not credited to it. 



Use of "Obligations" for "Expenditures" 



Although the study is labeled Federal Expenditures 

 for Research and the word "expenditures" is used in 

 most of the tables as well as in the text, "obligations" 

 were actually used so far as possible. Use of "obliga- 

 tions"' aids in classifying "expenditures", since data by 

 separate projects are often available on that basis in the 

 "green sheets" submitted by agency budget officers to 

 the Bureau of the Budget each September 15 and later 

 published in the Budget of the United States Govern- 

 ■irient. However, when more feasible, use was made of 

 amounts on other bases — "checks issued", "unrevised 

 cash", "allocations" of administrative funds, or even 

 "appropriations" and (for a few works projects) "au- 

 thorizations". In most cases, the differences between 

 the amounts reported on the several bases would be 

 smaller than the changes that would result from either 

 an altered definition of "research" or from different 

 methods of applying the necessarily intangible concept 

 of "research" to specific expenditure items. 



The use of "obligations" rather than "expenditures" 

 affects somewhat the inclusion of "research" work in 

 those few agencies that have such work transferred to 

 them, or carried on by another agency for them. This 

 study has followed, with one exception, the Govern- 

 ment's accounting and budget procedure in the case 

 of such transfers, charging "obligations" to the agency 

 actually performing the technical work, rather than 

 to the agency that eventually pays for the operation 

 and expects to use the results. Since the agency car- 

 rying on the work can give a break-down of the costs 

 while the reimbursing agency knows that only certain 

 work including some research is billed to them, this 

 method is more feasible in trying to separate costs 

 of projects. The determining factor for this study, 

 however, was the better information about research 

 projects available on the "obligations" basis. 



Definition of "Agency" 



For purposes of this study, "agency" was defined to 

 mean either (1) an independent establislunent or (2), 



87 



in the 10 Cabinet Departments, any bureau, office, 

 division, administration, service, college, survey, or 

 other major unit the head of which reports directly 

 to the Secretary, Under Secretary, or Assistant Sec- 

 retary of the Department without nominal routing 

 through an intermediate bureau chief or other admin- 

 isti'ative officer. 



Accuracy and Consistency 



Consistency rather than accuracy is the criterion of 

 excellence in a compilation of expenditures for a func- 

 tion such as research. The definition of "research" 

 used in this report is necessarily arbitrary. Again, 

 the methods of applying the definition to the different 

 types of work encountered must be arbitrarily deter- 

 mined to fit each individual case, especially in such 

 matters as allocating general administrative expenses, 

 travel, printing, etc., between reseaixh and nonresearch 

 activities. Hence accuracy, although necessarily im- 

 plied in any numerical compilation, is without its es- 

 sential meaning of conformity to an exact standard, 

 and consistency takes its place as a technical criterion. 

 The degree of consistency in the data can be judged, 

 however, only by considering feasible alternatives, and 

 hence is largely a matter of individual judgment. 



Data on the basic tables are carried to the nearest 

 thousand dollars, irrespective of size or of the probable 

 degree of care and understanding involved in indi- 

 vidual estimates. Most of the material, as received 

 from the agencies was expressed in units of hundreds 

 of dollars or less, although a few estimates were car- 

 ried only to the nearest ten thousand dollars. Several 

 agencies reporting less than $10,000 for research spent 

 an unusual amount of time trying to get what they 

 considered very accurate figures, in spite of the sug- 

 gestion that it was not necessary. 



The data on emergency obligations for research pur- 

 poses may be 10 percent or more in error for the fiscal 

 year 1937 and jjerhaps twice this much for 1938. The 

 probability is, however, that detailed complete com- 

 pilations of the emergency expenditures would not 

 change the results more than one or two million dol- 

 lars. The method of estimating the emergency ex- 

 penditures is explained in a later section of this 

 Appendix. 



Omissions 



Army and Navy salaries and equipment. — One of the 

 most important omissions, measured by quantity, in 

 data reported is an amoimt to represent salaries of 

 officers and enlisted personnel from the Army and 

 Navy Departments assigned to research projects, and 

 the cost of the regular Army and Navy equipment and 

 supplies used in making field tests of new devices or 

 ordnance. In each of these two Departments the pay 



