SECTION I. GENERAL REMARKS. 



In the present paper I have attempted a statistical exami- 

 nation of Anglo-Indian Stature based on Dr. Annandale* s records. 

 The measurements were all taken by Dr. Annandale or in a few 

 cases under his direct supervision. Thus the present material may 

 be considered free from large fluctuating errors due to different 

 personal bias of different observers. 



Nature of the Material. 



Dr. Annandale has explained in his introductory note the 

 special character of the present material. After excluding 

 " Negro," " West Indies" <( Chinese," " Burmese" and "Bhutia" 

 ancestr}* and omitting certain incomplete and doubtful records a 

 series of 200 was obtained for Stature, Head Length, Head Breadth, 

 Nasal Length, Nasal Breadth. Zygomatic Breadth and Upper 

 Face Length. 1 



The great importance of the present material from a biomedi- 

 cal standpoint will be easily appreciated. So far as I am aware 

 this is the first time that a true biologically mixed population is 

 being studied by statistical methods. 



From the statistical standpoint the coefficient of variability is 

 considered to be a very important test of homogeneity.' 2 Hitherto 

 all attempts to fix the upper limit of homogeneous variability were 

 necessarily confined to the study of artificially made up mixtures. 8 

 The Anglo-Indian data furnish us with a ''natural mixture." A 

 careful study may be expected to throw considerable light on this 

 vexed question. Incidentally, it will be of great interest to com- 

 pare the variability of such a "mixed" population with those of 

 " purer" races.* 



The Anglo-Indian population may really represent a new 

 ' race " in the making, and we hope to discuss in the sequel what. 

 indications may be afforded by a study of the present material as 

 regards the mechanism of race formation. 



It should be noted however that the word " race " is here used 

 in its statistical sense. Pearson 5 says, "Any race may originally 

 have arisen from a mixture of races, but such a mixed race is 

 wholly different from a mixture of races, which have not interbred." 



Arithmetical work on these characters is nearly finished and I hope to 

 I ublish the results ai an early date. 



rhis is trueoi course for uni-modal data only, or more generally for distribu- 

 tions which cannot be dissected into component frequency groups. For a fuller 

 discussion ol this point see pp. 34, 93-94. 



C. S Myers-*Man, February, [903, pp. 28-32. Also sec Karl Pearson's 

 discussions on this poinl in Biometrika Vol.2, 1903,00. 345-347, Myers' Reply 

 and 1 us Remarks on this Reply in Biometrika vol. 2, 1903, pp. 504-508. 



" Purer" in a statistical sense, i.e. more homogeneous. 

 h Biometrika Vol. 2, 1903, p. s ,if) . 



