September 9, 1921] 



SCIENCE 



223 



one of them possesses an entire organ that is 

 absent in the other. Sconiber scomhrus is 

 without a swim bladder; Scornber japonicus 

 has a well-developed one. 



This is a most glaring inconsistency. On 

 one hand, to separate two genera on the basis 

 of a mere modification of an organ that is 

 possessed by both of them, and on the other 

 hand, to include in one genus two forms, one 

 of which possesses an organ that is absent in 

 the other. Making this inconsistency more 

 marked is the fact that in the case of the 

 sharks it is only during a part of the life of 

 the animals (when they are with young) that 

 the character of the ' placenta,' upon which 

 the genus is based, can be ascertained. In 

 the mackerel the presence or absence of the 

 swim bladder can be seen at any time by 

 simply opening the abdominal cavity. 



On the whole, workers in vertebrate tax- 

 onomy seem to be more chary than those in 

 invertebrate, in making use of internal charac- 

 ters in classification. The fact that a character 

 is not readily apparent should not influence 

 its use if animals are to be arranged in their 

 true relationship. 



Such a marked structural difference as the 

 IKDSsession of an organ as compared with the 

 suppression of it certainly should be consid- 

 ered of generic weight. Therefore it would 

 seem well to raise the subgenus Pneumato- 

 phorus Jordan and Gilbert, to generic rank. 

 The American species. Scomber colias and 

 S. japonicus, would thus stand Pneumato- 

 pJiorus colias and P. japonicus. 



Edwin C. Starks 



an improved method of estimating the 

 number of genetic factors con- 

 cerned in cases of blend- 

 ing inheritance 



Dr. Sewall "Wright has kindly pointed out 

 an error in the formula which I recently sug- 

 gested^ in connection with this subject. In- 

 stead of taking the direct difference between 

 the standard deviations of F^^ and F„, as I 

 did, one should deal with the difference be- 

 tween the squared standard deviations. Dr. 

 Wright bases this correction on his discussion 



1 Science, July 29, 1921. 



of the fundamentals of factorial theory as de- 

 veloped particularly in " Systems of Mating 

 IV.," Genetics, 6, March, 1921. He gives the 

 correct formula for the number of factors (.n) 

 concerned in a case of blending inheritance as 



_ Dj 



in which D is the difference between the means 

 of the parental races, o-j is the standard devi- 

 ation of -Pj, and (To is the standard deviation 

 of F^. This method gives in general a smaller 

 number of genetic factors than the method 

 which I suggested, and its use is simpler. 

 Applied to the examples which I cited, it 

 gives, in the ease of seed weight of maize, 4 

 or 5 factors instead of " about 15 " ; and in 

 the case of weight of rabbits in three different 

 crosses, 3, 14 and 22 or 23 factors, instead of 

 66, 80, and 176, respectively. I am greatly 

 indebted to Dr. Wright for the correction. 

 W. E. Castle 



THE CURVE OF DISTRIBUTION 



To THE Editor of Science: An explanation 

 of the irregularities in the curve of the dis- 

 tribution of the heights of 221,819 men, taken 

 from insurance statistics, to which Professor 

 Boring called attention in Science for Novem- 

 ber 12, 1920, may possibly be found in the 

 nature of the measuring devices used by the 

 examining physicians. One of the three lead- 

 ing types on the market and at least one other 

 are graduated in inches alone instead of in 

 feet and inches. The tendency for men who 

 use these scales to read off the round num- 

 ber, 70 inches, instead of 69, and 60 inches 

 instead of 59, might be great enough to ac- 

 count for the " bumps " in the Gaussian curve 

 at 5 ft. 10 in. and at 5 ft.; and the lowering 

 of the average height which would result from 

 the correction of these exaggerations might 

 change the ideal curve sufficiently to bring 

 the bump at 5 ft. 8 in. within the normal limits 

 of error for a curve whose unit of measurement 

 is so large in comparison to the total range of 

 variation. 



Carl H. P. Thurston 



Pasadena, Calif. 



