714 



SCIENCE 



[N. S. Vol. XXXVI. No. 934 



sarily a "highly speculative discipline," but 

 rather a field for quantitative research. Weight 

 may be given to this statement by a brief de- 

 scription of an experiment made this year at 

 the Station for Experimental Evolution. 



Much of the biometric work on selective mor- 

 tality has necessarily been of a highly statis- 

 tical character, but this particular experiment 

 has the virtue of extreme simplicity. In the 

 spring of 1912, a series of about 238,000 bean 

 seedlings veas examined for morphological 

 variations to serve as a basis for experiments 

 in selection within the " pure line." Of these, 

 about" 4,217 abnormal' and 5,030 normal* seed- 

 lings were transplanted to the field. In doing 

 this great care was used to maintain precisely 

 comparable conditions for both normal and 

 abnormal plants. As plants died, from any 

 cause" whatever, their labels were brought in 

 and at harvest time a summary was prepared 

 showing the numbers of seedlings failing to 

 develop to fertile maturity. 



Of the 5,030 normal plants, 226, or 4.493 per 

 cent., died. Of the 4,217 seedlings showing 

 some morphological variation from type, 286, 

 or 6.782 per cent., failed to reach maturity. 



' ' The Measurement of Natural Selection, ' ' ap- 

 pearing in The Popular Science Monthly, Vol. 78, 

 pp. 621-638, 1911. Several other studies have 

 been published since the writing of that r^sumg. 



' The numbers given here are substantially cor- 

 rect, but may be slightly modified when the records 

 are verified by cheeking against the labels of the 

 individual plants. This can not conveniently be 

 done until the 8,000 and more individually wrapped 

 plants are opened for shelling and planting in the 

 spring of 1913. 



'Abnormal includes all morphological devia- 

 tions from the normal type. 



' For every abnormal seedling found at least 

 one normal was taken quite at random from the 

 same seed flat. The chief reason for the excess of 

 normals is that in some lines the quantity of seed 

 was not as large as necessary for securing a good 

 number of abnormals, and in these cases normals 

 were planted to avoid losing the line. 



"An exception is made in the case of a large 

 area of plants which were completely ruined when 

 nearly ripe by obviously non-selective causes out- 

 side the experimenter's control. 



Line 



1- 10 



11- 20 



21- 30 



31- 40 



41- 50 



51- 60 



61- 70 



71- 80 



81- 90 



91-100 



101-110 



111-120 



121-130 



131-140 



141-150 



151-160 



Thus under conditions of careful cultivation, 

 with ample space, with no intra-specific and 

 practically no inter-specific competition, and 

 with a general mortality of less than 5.55 per 

 cent, there is a clearly marked selective death 

 rate. 



Now if p be the number which perish in a 

 population of m individuals the probable error 

 of that number is given" by 



£', = .67749^px(l-|) 



From the absolute probable error, the percent- 

 age probable error is at once obtained by tak- 

 ing the ratio of 100 p to m. Thus we have for 

 the death rates: 



Per Cent. 



For normals 4.49 ± .20 



For abnormals 6.78 ± .26 



Difference 2.29 ± .33 



Thus the difference is seven times its prob- 

 able error, and is clearly trustworthy statis- 

 tically. That it is not due to chance is most 

 strongly brought out by splitting the material 

 up into 16 lots of about ten " pure lines " each, 

 and determining the death rate for normals 

 and abnormals in each lot separately. The 

 little table gives the results. 



Because of the low mortality great irregu- 

 larity is to be expected in the results. But in 



"Biometrilca, 2: 274, 1903. 



