Sbptembee 5, 1913] 



SCIENCE 



345 



tivity of electrolytes in a mixture of solvents, 

 and conductivity of a mixture of electrolytes 

 in pure solvents. The tables of electromotive 

 forces include those of normal cells, of transi- 

 tion cells, of concentration cells, the potential 

 of simple electrodes, and divers unclassified 

 electromotive force effects. 



In addition to the above there are forty- 

 seven pages devoted to data in general elec- 

 tricity and magnetism. Immediately follow- 

 ing these are eight pages on radioactivity and 

 ionization. The vsTiter finds nothing on the 

 Peltier effect or on the important subject of 

 electrolytic thermo-electromotive force. 



A bibliography is appended to every main 

 division of the book. An alphabetical index 

 would add much to the convenience of refer- 

 ence. The second volume for 1911 contains 

 both a general and a special alphabetical list 

 of all substances mentioned in both volumes. 



Henry S. Carhart 



SPECIAL ARTICLES 



AN ILLUSTRATION OF THE INFLUENCE OF 



SUBSTRATUM HETEROGENEITY UPON 



EXPERIMENTAL RESULTS 



In experimental breeding so much stress 

 has been laid upon controlled fertilization that 

 some other factors of importance for the ob- 

 taining of trustworthy results have been left 

 too much out of account. The importance of 

 heterogeneity in the substratum upon which 

 the plants are grown as a possible source of 

 error has been pointed out time and again. 

 De Vries, for example, attaches great weight 

 to this factor. 



The purpose of this note is to give point to 

 these warnings (too greatly neglected now) by 

 showing how extrinsic influences may com- 

 pletely screen intrinsic tendencies. 



In very extensive series of materials a posi- 

 tive correlation has been demonstrated between 

 the weight of the seed planted and the number 

 of pods on the plant into which it develops — 

 that is, yield is higher in the plants from the 

 heavier seeds. This is true without exception 

 for twenty series, involving 13,099 plants, 

 already published.' Further constants based 



on 4,856 plants, are given b«low. Here the 

 coefficient of correlation, r,„p, shows the rela- 

 tionship between the weight of the seed 

 planted (in the conventional units of .025 

 gram range) and number of pods per plant, 

 while the second term of the regression 

 straight line equation,^ 



(P — Tap — w) + Twp — w, 



shows the absolute change in number of pods 

 per plant for each unit change in seed weight. 



The constants are in excellent agreement 

 with those already published — fairly large and 

 positive throughout — with the exception of 

 the Golden Wax, the L series, and the GG 

 culture of Burpee's Stringless. Those for the 

 Golden Wax series, LG and LL, are sensibly 

 zero ; one is the smallest positive coefficient yet 

 found while the other is negative in sign, 

 though only a fraction of its probable error. 



The coefficient for the GG series is in strik- 

 ing contrast to the others; not only is it 

 numerically the largest value recorded, but it 

 is negative in sign and unquestionably signif- 



' Harris, J. Arthur, ' ' The Eelationship between 

 the Weight of the Seed Planted and the Char- 

 acteristics of the Plant Produced — I.," Biomet- 

 rika, Vol. 9, pp. 11-21. See also Amer. Breed. 

 Mag., Vol. 3, pp. 293-295. 



-p = pods per plant, «)=weight of seed planted. 

 The bars indicate the means and the sigmas denote 

 the standard deviations of the characters in ques- 

 tion. Through a slip in the copying of the manu- 

 script which I overlooked in the proofs, the second 

 term of the regression formula is given the nega- 

 tive sign on p. 14, BiometriJca, Vol. 9. The 

 values in the calculated equations are of course 

 correct. 



