184 GROVE KARL GILBERT— DAVIS [MEM0IRS [ v„^Txt 



any concrete result to which the discussion led. The treatment recalls that of the address on 

 the "Inculcation of scientific method by example," delivered 10 years before, but differs in 

 placing greater emphasis on certain steps in the orderly progress of a research. Its outline is 

 as follows: 



The first step in an investigation, after observation has been somewhat advanced, is the 

 invention of explanatory hypotheses. Hypotheses ordinarily "flash into consciousness with- 

 out premonition," and they appear always to be suggested through analogy. 



The unexplained phenomenon on which the student fixes his attention resembles in some of its features 

 another phenomenon of which the explanation is known. Analogic reasoning suggests that the desired explana- 

 tion is similar in character to the known, and this suggestion constitutes the production of a hypothesis. 



GUbert said that, in order to test this view, he had for some years endeavored to analyze 

 the methods employed by himself and some of his associates in geologic research. 



Next, after an investigator has framed or invented an hypothesis, he proceeds to test it. 

 If the phenomenon was really produced in the hypothetic manner, then it should possess, in 

 addition to the features already observed, certain other specific features, and the discovery of 

 these wdl serve to verify the hypothesis. Resuming its examination, he searches for the par- 

 ticular features. If they are found the theory is supported; and in case the features thus pre- 

 dicted and discovered are numerous and varied, the theory is accepted as satisfactory; but if 

 the reexamination reveals features inconsistent with the tentative theory, the theory is thereby 

 discredited, and the investigator proceeds to frame and test a new one. The investigator there- 

 fore makes two steps: He invents hypotheses and he invents tests for them, and it was to the 

 intellectual character of these two kinds of invention that attention was directed. The com- 

 ment may be here introduced that tests are usually prepared by the largely Conscious process 

 of deduction, and thus differ from hypotheses, which are more truly inventions. 



The treatment then turns more especially to Coon Butte and considers particularly 

 the two hypotheses of meteoric impact and subterranean explosion, appropriate tests being 

 developed for each. The meteoric hypothesis is first examined. 



If the crater were produced by explosion the material contained in the rim, being identical with that removed 

 from the hollow, is of equal amount; but if a star entered the hole, the hole was partly filled thereby, and the 

 remaining hollow must be less in volume than the rim. 



The excess of rim matter required by the theory of a "buried star" was not found by 

 observation; nor were any magnetic deviations detected, such as a large meteor should produce. 

 It was calculated that if the meteor were 500 feet in diameter, it must be buried 10 miles below 

 the surface in order not to affect the magnetic needle perceptibly; and such a depth of burial is 

 regarded as so improbable that the meteoric hypothesis is thereby negatived and excluded. 

 The hypothesis of a subterranean explosion, caused by an ascending volcanic intrusion nearly 

 to the surface, is next examined and found to be more plausible; but in spite of the opinion 

 earlier expressed in favor of this hypothesis, little preference is expressed for it over several 

 other hypotheses which are patiently examined. In the end, the origin of Coon Butte is left 

 uncertain, as if both the hearers and the readers of this address were urged to practice the sus- 

 pension of judgment until some more convincing test than any that Gilbert had devised could 

 be tried and satisfied. Instead of emphasizing one or another explanation of the curious 

 topographic feature to which so much careful observation and reflection had been given, 

 emphasis was given to the conditions under which an investigator may best invent hypotheses. 

 It was concluded that if the ideas set forth concerning the origin of hypotheses are correct, 

 "then fertility of invention implies a wide and varied knowledge of the causes of things, and 

 the understanding of nature in many of her varied aspects is an essential part of the intellectual 

 equipment of the investigator." 



Gilbert was undoubtedly right in his estimate of the more and the less important aspects 

 of his address. Coon Butte is a relatively trifling and local affair, while fertility in the invention 

 of hypotheses and of tests by which the hypotheses can be proved or disproved is a matter of 

 the broadest scientific importance. But if, in the judgment of a man in Gilbert's position, it 

 was desirable in 1895 to expound to a representative body of American specialists in the hightly 



