236 ■ Memoirs Bcniicc P. Bishop Mitsctiin 



point out in some detail just wliat has been attempted in describing characters that 

 do not lend themselves readily to measurement. The fact that anthropologists have 

 carelessly spoken of "types" of hair form, hair color, or eye color has given the 

 erroneous impression — not only to the general reader but to many anthropologists 

 as well — that these types actually exist in nature and that it is possible, for 

 example, to arrange all human eyes in four, five, or six color groups. Although it 

 is universally recognized that all characters that lend themselves to actual meas- 

 urement show a continuous variation with a tendency for a large percentage of the 

 individuals measured to cluster around a median or mean point, yet it is difficult 

 to dislodge the idea that other characters such as color or form, which cannot be 

 accurately measured with existing apparatus, have a discontinuous distribution. 

 The body height or stature of the Scots, for examiile, ranges from 158 centimeters 

 to 186 centimeters and averages about 172 centimeters. \"ery few Scotchmen are 

 as short as 158 centimeters and very few are as tall as 186 centimeters. In pro- 

 gressing from the extremes towards the mean the number of individuals at each 

 step increases, it is apparent to anviine who has endeavored to classify characters 

 which do not lend themselves to measurement that in them he is dealing with 

 exactly the same type of continuous variation. When the metric rod cannot be 

 applied, standards are set up along the range of variation, separated widely enough 

 to permit of distinguishing each from the standard preceding or following it, and 

 an endeavor is made to classify the material on this basis. The attempt to classify 

 all existing forms of hair as straight, low waves, deep waves, curly, frizzly, or 

 woolly, produces results ^'ery similar to those which might be expected if the stature 

 of all men were measured with a rod graduated in 10 centimeter intervals from 

 130 to I go centimeters. A man's stature would be recorded as 130, 140, 150, 160, 

 170, 180. or 190 centimeters; yet it is obvious that the stature of manv men would 

 actually be 135, 136, or 137 centimeters. The rod is not graduated finely enough 

 to record the true distribution of the measurements. In a sense hair classes may 

 be compared with these lo-centimeter intervals. For example, straight hair might 

 well correspond with the 130-centimeter mark and woolly hair with the lyo-centi- 

 meter mark or vice versa. But at this ]X)int the analogv breaks down. It is not 

 certain that low waves, deej) waves, and other hair forms correspond exactlv to the 

 140- and 150-centimeter points. Roughl\- the}' probably do. But by far the great- 

 est difference in the two methods and one that should always be kept in mind in 

 the analysis of data is that in the classification of these descriptive or attribute 

 characters, so called, the "metric rod" exists only in the mind of the observer and 

 is by no means a uniform or universal standard. This lack of a fixed standard 

 makes difificult not only the comparison of small differences found b\' different 

 observers, but also to a lesser extent those found at different times by the same 

 observer. As the standard is purely visual, constructed largely upon the expe- 



[6] 



