G. K. Gilbert—Special Processes of Research. 455 
was a certain amount of antecedent knowledge with respect to 
the subject of investigation, and this determined a large portion 
of the scheme; thus the nearness of the stations was dictated 
by approximate knowledge of the size of the storms to be ob- 
served. In the second piace, certain hypotheses were enter- 
tained in regard to the phenomena, and these had their in- 
fluence; for example, it was tentatively postulated that the 
beginning of rainfall bore such a constant relation to the storm 
as a whole that the timing of it would afford a record of the 
progress of the storm. In the third place, analogy with other 
researches had its influence; a certain resemblance was recog- 
nized between thunderstorms and cyclonic storms, and the gen- 
eral system of investigation was modeled upon that which had. 
proved successful in the study of the larger meteors. Finally, 
numerous features of the system were determined by what may 
be regarded as obstacles to observation. The limited endow- 
ment of the investigation was such an obstacle and occasioned 
the restriction of observation to a certain season, the further 
‘restriction to the actual period of storm occurrence, and also 
the classification of observers. 
This classification of considerations controlling the organiza- 
tion of the research is believed to be comprehensive, and appli- 
cable to any other systematic research, but it will not serve the 
purpose of dividing into groups the details of working plan to 
which the considerations give rise, because nearly every 
detail is the joint product of considerations falling in two or 
more categories; and it should be observed also that its 
categories are not mutually exclusive. The consideration of 
obstacles—or practical limitations—never occurs alone, but 
merely modifies the procedure indicated by other considera- 
tions. The consideratio’ of analogy cannot stand alone, be- 
cause analogy exists only in virtue of some knowledge or sup- 
posed knowledge or hypothesis with reference to the nature of 
the subject of research. And again, considerations of ante- 
cedent knowledge and of hypothesis have no clear line of 
demarcation, for the practical discrimination of knowledge and 
hypothesis, however sharply they may be divided in thought, 
is beset with insuperable difficulties. 
Returning to our illustration, we note that the published 
plan of observation includes no mention of barometric pressure ; 
and that the only observations touching the electric condition 
of the atmosphere are those on lightning and thunder. It is 
easy to understand that these omissions are referable to the 
obstacle of expense, and we may also assume, although we 
have no warrant in the publications of the society, that such 
observations will be made at some future stage of the investiga- 
tion, when accumulation of knowledge and hypothesis has 
e 
