SCIENCE (and common SENSE ) 31 



now in much the way that "similar things" behaved there and then, 

 and will again behave in other eras and places. 



Dissolubility. Believing in the continuity of nature, I do not con- 

 sider it an absolute continuum. Acknowledging myself a part of na- 

 ture, I still consider myself separate from it. More significantly, we 

 all conceive ours a world of distinct, if interacting, "objects" and 

 "phenomena," in particular "situations" largely independent of the 

 rest of the external world. All our labors of prediction assume this 

 kind of dissolubility, assume each case measurably affected by a 

 relatively small number of conditions— each state of a system ade- 

 quately defined by a relatively small number of variables. Prediction 

 would be impossible had we first to specify or determine the state of 

 the entire universe. Confidently seeking colligative relations, com- 

 mon sense presupposes a principle of dissolubility. Seeking laws, so 

 does science. 



Emboldened by our predictive successes, both in science and in 

 common sense, we make of the ideas of determinism, continuity, and 

 dissolubility general principles that transcend all possibility of em- 

 pirical justification. These "metaphysical" principles, together with 

 a few others more peculiar to science, are examined in detail in 

 Chapter VII. I broach them here because the sharing of these prin- 

 ciples produces in both science and common sense certain shared 

 criteria for the selection of "fit" subject matter. 



The Subject Matter of Science 



We expect recurrence in a world to which the principle of continuity 

 applies, regularity in a world to which the principle of determinism 

 applies, agreement of eyewitnesses in reports of an existing "external 

 world." The principles we accept thus determine the selected ele- 

 ments of experience alone acceptable to us as facts. We are prepared 

 to accept as facts only those elements of experience that are actually 

 or potentially reproducible, regidar enough to offer promise of de- 

 terminist order, and concurrently reported by all "normal" observers. 



RECURRENCE AND REPRODUCIBILITY 



Guided by the principle of determinism, we expect that, in circum- 

 stances of adequate control, at least approximate reproduction of the 

 conditions must yield at least approximate reproducibility of tlie 



