January 3, 19 18] 



NATURE 



355 



THE PRODUCTION OF SCIENTIFIC 

 KNOWLEDGE.^ 



THE increase of scientific knowledge can be divided 

 into three steps : first, the production of new 

 knowledge by means of laboratory research; secondly, 

 the publication of this knowledge in the form of 

 papers and abstracts of papers j thirdly, the digestion 

 of the new knowledge and its absorption into the 

 general mass of information by critical comparison 

 with other experiments on the same or similar sub- 

 jects. The whole process, in fact, may be likened to 

 the process of thought. We have first the perception 

 by means of the senses. The percept is then stored 

 in the memory, and in the mind is compared with 

 other previously stored percepts, and finally forms with 

 them a conception. 



I desire in this paper to consider the methods by 

 which these three sections of the production of know- 

 ledge may be carried on, to suggest an arrangement 

 of laboratories to produce experimental results dealing 

 with any branch of science, then to consider how the 

 knowledge so obtained may best be stored and classi- 

 fied, and, finally, the methods to be employed to make 

 the results of scientific research available for applica- 

 tion. 



(i) Research Work. 



The agencies engaged in scientific research are of 

 several kinds. The traditional home of research work 

 is in the university, and the bulk of the scientific 

 production of the world comes from institutions con- 

 nected with teaching. The industries are more and 

 more supporting research laboratories, a large number 

 of which contribute to the general fund of scientific 

 knowledge by publishing the results which they obtain, 

 and some of which are engaged upon purely scientific 

 work of no mean order. Consulting and technical 

 laboratories engaged in industrial work make frequent 

 contributions to science, and there are some very 

 important laboratories engaged in pure research work 

 which are supported by philanthropic foundations. 



The classification of research laboratories is not alto- 

 gether an easy task. They may obviously be classified 

 according to the source of the funds which support 

 them — that is, we may classify them as university 

 laboratories, industrial laboratories, Government 

 laboratories, institution laboratories, and so on — 'but if 

 we look at them simply in the light of the research 

 undertaken, this does not seem to be altogether a 

 logical classification, since there is little distinction 

 between the work done in some university laboratories 

 and some industrial laboratories, and the work of the 

 Government and institution laboratories again overlaps 

 that of the two former classes. 



The University of Pittsburg, for instance, has an 

 industrial laboratory, where definitely technical 

 problems are dealt with. The research work on photo- 

 metry done at Nela Park and at Cornell University 

 would seem to be similar in kind, and work on physical 

 chemistry or on the structure of chemical compounds 

 is of the same type, requires the same class of workers, 

 and produces the same results, whether it be done in 

 a university, in a laboiatory of the Carnegie Institu- 

 tion, or in such an industrial laboratory as that of 

 the General Electric Co. It is equally difficult to 

 classify laboratories according to the purpose for which 

 researches are avowedly carried on. Most universitv 

 laboratories are willing to undertake work of industrial 

 value, and, indeed, some specialise in such problems, 

 while many industrial laboratories are quite willing 

 to carry out a research of purely academic and theo- 

 retical interest provided the problems involved bear 

 a relation to the general work of the laboratory. 



1 From a paner re.id before the Rochester Sertion of the Optical Society 

 of America on October 23, by Dr. C. E. Kenneth Mees. 



NO. 2514, VOL. I go] 



i A useful classification of laboratories can, however, 



j be obtained if we consider whether the problems in- 



I vestigated in a laboratory are all connected with one 



I common subject or whether the problems are of many 



kinds, having no connecting bond of interest. I would 



suggest that the first type of laboratory might be 



called "convergent" laboratories, and the second 



"divergent." 



In the "divergent" group of laboratories are in- 

 cluded all those institutions where research is carried 

 on which are interested in science in general or in 

 science- as applied to industry, and will attack any 

 problem that may seem to promise progress in 

 knowledge or, in the case of an industrial laboratory, 

 financial return. Most university laboratories are of 

 this type. When they devote themselves to special 

 problems it is usually because of the predilection of 

 some professor, and as a general rule a student or 

 instructor may choose any problem in the whole field 

 of the science in which he is working and may carry 

 out an investigation on that problem if he be interested 

 in it without regard to the relation of his work to 

 the other work which is carried on in the same labora- 

 tory. 



Correspondingly, in iViost industrial laboratories the 

 problems investigated are those which present them- 

 selves as a result of factory experiences or of sugges- 

 tions from the men working in the laboratory, and 

 promise financial return, and the different problems 

 carried on in the same laboratory are not necessarily 

 related in any way whatever. 



The greater number of university and industrial 

 laboratories are necessarily of this type. It would be 

 a disadvantage for a university laboratory, the primary 

 business of which is training students, to be too nar- 

 rowly specialised. Specialised university laboratories 

 are desirable only in the case of post-graduate students, 

 and it would be very inadvisable to allow the labora- 

 tories responsible for the general training of scientific 

 men to specialise in one branch of science, since as a 

 result the students would acquire a proper acquaint- 

 ance with only a limited portion of their subject. 



Industrial laboratories, on the other hand, must 

 necessarily be prepared to deal with any problems pre- 

 sented by the works, and as these -will be of all kinds, 

 covering generally the whole field of physics, chem- 

 istry, and engineering, it is impossible for the usual 

 works laboratory to sp«cialise except in so far as it 

 deals with the works processes themselves. 



In the " convergent " laboratories, however, although 

 the actual investigations may cover as great a range 

 of science as those undertaken in a "divergent" 

 laboratory, yet all those investigations are directed 

 towards a common end — that is, towards the elucidation 

 of Associated problems related to one subject. Thus, 

 the staff of the Geophysical Laboratory, which includes 

 physicists, geologists, crystallographers, mineralogists, 

 and chemises, works on the structure of the rocks, and 

 although the field of the actual investigations ranges 

 from high-temperatu-e photometry to the physical 

 chemistry of the phase rule, yet the results of all the 

 work carried out are converged on the problem of the 

 structure of the earth's crust. 

 I The Nela Park Laboratory, in the same way, is 

 I studying the production, distribution, and measurement 

 i of illumination, and all its work, which may involve 

 I physiology, physics, and chemistry, is relatei to that 

 ; one subject. Such convergent laboratories sometimes 

 develop in universities owing to the intense interest of 

 a professor in a single subject and to the enthusiasm 

 which inspires students and assistants to collaborate 

 with him and to concentrate all their energies on the 

 same group of problems. There are many examples of 

 such laboratories, such as the laboratories dealing with 

 radio-activity, and those which are concerned chiefly 



