2 50 ANDERSON 



has found his way into it. The finicky pointer-reading data, single sense im- 

 pressions, lengths, widths, weights, so useful for precise analysis, are best 

 deferred until we know what kind of a problem we are up against. 



Pattern data have a broader observational basis than pointer-reading data, 

 as Minot pointed out (1911) half a century ago in his pioneer attempt to 

 fit growth curves. Therefore they are invaluable in the early stages of any 

 complex problem. Before we have some notion of what we are about, pointer 

 readings by themselves are little help. Lengths, densities, weights, though 

 accurately determined, merely allow us to phrase our ignorance more ele- 

 gantly. Precision has little advantage until we have enough understanding to 

 use precise analysis. In the early stages of a problem, accurate, unrelated 

 data, if collected in big enough quantities by many people and scattered 

 through numerous papers, may actually obscure the problem and hinder its 

 solution. Platoons of busy workers may share their data so expertly that only 

 a gifted few realize the fundamental obscurity of the problem. I remember 

 with a shudder the lectures and the seminars on growth tropisms I endured 

 before the Wents (following Darwin's lead) cracked the whole problem wide 

 open by an elegant combination of pattern data (curvature of Avena coleop- 

 tiles) and pointer readings (analysis of the growth substances concentrated 

 in the agar blocks). 



The most rapid scientific advances come when some genius recognizes a 

 pattern which is diagnostic, which can be used with precision, and which 

 orders up whole pages of pointer readings. Such was the recognition of the 

 periodic system, such was Mendel's lighting upon the Mendelian factor as 

 a hereditary unit. He noted that with the round versus the wrinkled peas, 

 with the tall versus the short and fasciated plants, he had significant patterns 

 which behaved as units in their transmission from generation to generation. 

 We do not yet know the precise relationship between these unit patterns and 

 the germ plasms of which they are a part, but the recognition that they were 

 more or less the same sort of thing and that they were significant patterns 

 with which to analyze problems of inheritance has made Genetics one of the 

 most fertile fields in biology during the past fifty years. 



Those who have not actually participated in such routine genetic chores 

 as the making of linkage maps seldom realize how qualitative the basic data 

 of Genetics can be. It is a simple point, but worth emphasizing. Genetics 

 advanced so rapidly because it was using pattern data with their broad 

 observational basis, but using them with great precision. Take the eye mu- 

 tants in Drosophila for instance. In an experiment with several marker genes 

 involving eye color, one could catalogue them and their various recombina- 

 tions efficiently only when he had learned to recognize as significant patterns 

 various little differences in color, texture, size, and aging. Imagine trying to 

 separate two wing mutants of Drosophila on a truly objective, statistical 



