PHYSICAL AND BIOLOGICAL THEORIES 51 



behaviour of gases under pressure with Darwin's theory of natural 

 selection. In the former case much thought was needed to gather 



ic data ; but they could be exactly measured, and, once obtained, 

 single induction sufficed to enable the discoverer to perceive 

 lis law ; and, though a great deal of testing was still necessary, yet 



ie means of performing it were not hard to think of. In the 



itter case, the facts lay patent but multitudinous under the eyes 

 >f every man. Individuals, as regards some of their characters at 

 my rate, could be measured with exactness ; but the measurements 

 >f no child agreed with those of its parents. The difficulty lay, 



lerefore, not so much in observing facts as in interpreting them 

 tracing hidden analogies, in seeing likenesses amid differences, 

 separating the relevant from the irrelevant, essentials from non- 

 essentials, in brushing aside the veil of familiarity, in discarding 

 superstitions, in devising tests. Boyle's difficulties, however 

 ;at, were not of the same order as Darwin's. They lay 



lainly in discovering the facts, all of which were obscured. 



larwin's difficulties lay in tracking his way through the enormous 



id confusing maze of patent facts that spread and grew around 

 nm. 



81. Obviously biology does not lend itself to experiment so 

 idily as physics and chemistry. Apart from the circumstance 



lat many of its facts are patent, it is too complex. The first 

 iterpretation of a physical or chemical experiment is often 

 iccepted without cavil, and is comparatively rarely overthrown. 

 >ut seldom is the first interpretation of a biological experiment 

 xepted without hot controversy, and it is quite usual for the 

 >riginal interpretation to be overthrown and replaced by another, 

 tnd another, and yet another. These difficulties of interpre- 

 ition indicate the difficulties of satisfying the conditions 

 rhich make experiment useful the difficulty of so eliminating 

 relevant complications, that in the case of each experiment 

 ily one interpretation of the result is possible, or at least 

 jasonable. 



82. The notion that deduction is necessarily illegitimate and 

 langerous is demonstrably incorrect. That opinion has never guided 



successful students in their efforts to interpret nature. It is illegiti- 

 mate only when we start with an unproved assumption, or when it is 

 used merely to expand untested inductions by thinking which cannot 

 itself be tested by an appeal to reality. Rightly used, it is our 

 prime means of securing accuracy when thinking about sequences of 

 events. Physicists have constantly used it, and with magnificent 



