1888.] on Personal Identification and Descrij^tion. 355 



the radius of tlie circle of curvature of the depression at B, also of 

 that between the nose and the lip, for they are both very variable and 

 very distinctive. So is the general slope of the base of the nose. 

 The difficulty lies not in selecting a few measures that will go far 

 towards negatively identifying a face, but in selecting the best — 

 namely, those that can be most precisely determined, are most inde- 

 pendent of each other, most variable, and most expressive of the 

 general form of the profile. I have tried many different sets, and 

 found all to be more or less efficient, but have not yet decided to my 

 own satisfaction which to adopt. 



We will now suppose that either by the above method or by any 

 other, a standard collection of doubly outlined portraits such as that 

 in Fig. 3, has been made and come into use, so that a profile can be 

 approximately described by referring it to number so-and-so in the 

 catalogue. If the number it contained was less than 1000, three 

 figures would suffice to define any one of them. We will now con- 

 sider how a yet closer description of the jDrofile may be given by 

 using a few additional figures. One way of doing so is to have short 

 cross-lines drawn at critical positions between the two outlines of the 

 standard, and to suppose each of them to be divided into eight equal 

 parts. The intersection of the cross-lines with the outer border 

 would count as ; that with the inner border as 8, and the inter- 

 mediate divisions from 1 to 7. As the cross-lines would be very 

 short, a single numeral would thus defiiie the position of a point in 

 any one of them, with perhaps as much precision as the naked eye 

 could utilise. By employing as many figures as there are cross-lines 

 in the standard, each successive figure for each successive cross-line, 

 a corresponding number of points in the profile would be fixed with 

 great accuracy. Suppose a total of nine figures to be allowed, then 

 the first three figures would specify the catalogue number of the 

 portrait to be referred to, and the remaining six figures would 

 determine six points in the outline of the j)ortrait with greatly 

 increased precision. 



I may say that after numerous trials of different methods for 

 comparing portraits successively by the eye, I have found none so 

 handy and generally efficient as a double-image prism, which I 

 largely used in my earlier attempts in making composite portraits. 



I have not succeeded in contriving an instrument that shall 

 directly compare a given profile with those in a standard collection, 

 and which shall at the same time act with anything like the simplicity 

 of the mechanical selector, and with the same quick decision in 

 acceptance or rejection. Still, I recognise some waste of opportunity 

 in not utilising the power of varying the depths of the notches in the 

 cards, independently of their longitudinal position. 



Personal characteristics exist in much more minute particulars than 

 those just described. Leaving aside microscopic peculiarities, which 

 are of unknown multitudes, such as might be studied in the 800,000,000 



