October 17, 1907] 



NA TURE 



617 



either (a) a characteristic absorption, consisting of one or 

 more narrow bands and depending on the chemical struc- 

 ture of the dye molecule, or (6) a " resonance " spectrum 

 due to colloidal particles, and much more remotely con- 

 nected with chemical constitution. This spectrum is ill- 

 defined. 



The detailed e.xperiments which have led to these con- 

 clusions will be communicated in a paper shortly to be 

 published. A word or two is desirable in explanation of 

 the term ** resonance " spectrum. By this is denoted the 

 type of absorption exhibited by colloidal metal solutions, 

 glasses, and certain photographically prepared films 

 (F. Kirschner, DruUe's Atinalcn, 1904, -xiii., 239; Kirschner 

 and R. Zsigmondy, ibid., 1904, xv., 573 ; K. Schaum and 

 E. Schloemann, Zcit. wiss. Phot., 1907, v., 109). It is 

 probable that all absorption is due to resonance, no doubt, 

 and hence the narrow-band type (a), but in this case the 

 resonators would be the molecules or the contained 

 electrons, whereas in case (6) the resonance of larger 

 aggregates is the cause of the absorption. 



The investigation is to be continued, but a striking case 

 was found in one of the pinacyanols, a class of dyes 

 recently introduced as photographic sensitisers. This dye 

 gives in aqueous solution a flatfish, ill-defined absorption, 

 the solution showing all the characteristics of a colloidal 

 solution. In alcohol and organic solvents the absorption 

 was of a narrow-band type, entirely different, and this 

 spectrum was also obtained by heating the aqueous solu- 

 tion to boiling point. The behaviour was quite analogous 

 to that of starch, which gives crystalloid solutions at the 

 boiling point. S. E. Sheppard. 



Phys. Institute, Marburg a. ,L. 



The Convection Explanation of Electrolysis. 



At p. 12 of a recent text-book entitled " Electro- 

 chemistry," by Prof. R. A. Lehfeldt, the author mentions 

 the convection explanation of electrolysis, and states that 

 " Faraday was sufficiently impressed with it to form the 

 hypothesis of ions, i.e. of charged particles in the liquid 

 travelling under the action of the electric force." 



So far from being impressed favourably by this " explan- 

 ation," Faraday considered it might have a "dangerous 

 influence " and '* do great injury to science by contracting 

 and limiting .the views of those engaged in pursuing it." 

 He therefore constructed his terminology of electrolysis 

 specially to get clear away from this " explanation." 



Again, Faraday did not form a " hypothesis of ions as 

 particles." The word ion refers to the nature of the sub- 

 stance evolved at the electrode, not its dimensions. The 

 ion of Faraday might weigh an ounce or a ton. 



The opinions which the modesty of the great observer 

 permitted him to express may be found in the seventh 

 series of his "Experimental Researches." To attribute 

 to him- for any purpose, other opinions absolutely alien to 

 these is, I submit, either scientifically reprehensible or 

 grossly careless. J. Brown. 



Belfast, August 18. 



CLASSIFICATION OF PORTRAITS. 



pXPERIMENT.S of various kinds that I have inade 

 J--' to define the facial peculiarities of persons, 

 families, and races by means of measurement led to 

 the following- results that seem worthy of publication. 

 The most elementary form of portrait will alone be 

 considered here, namely, the outline of the face from 

 brow to chin, as in a shadow or in a silhouette. It 

 contains no sharply defined points whence measure- 

 ments may be taken, but artificial ones can be deter- 

 mined vvith fair precision at the intersections 

 of tangents to specified curves. It will be shown 

 that it is easy to " lexiconise " portraits by ar- 

 ranging the measurements between a few pairs of 

 these points in numerical order, on the same principle 

 that words are lexiconised in dictionaries in alpha- 



NO. 1 98 I, VOL. 76] 



betical order, and to define facial peculiarities with 

 greater exactness than might have been expected. 



The individuality of a portrait lies more in the 

 relative positions of six cardinal features (see the 

 figures , below) than in the shapes of the lines that 

 connect them, so long as the general character of the 

 connecting lines is roughly indicated. A few standard 

 types, perhaps ten in all (though I prefer (o use more), 

 represent as many concave, convex, and sinuous varie- 

 ties of outline, between each specified pair of the 

 si.K cardinal points, as need to be noted. I may recur 

 to this in a future letter. 



This will be apparent to the reader's satisfaction if 

 he compares portraits under unfavourable conditions, 

 as through a blurring medium, or out of focus ; or, 

 again, if he substitutes connecting links that differ 

 somewhat from the true ones. Consequently my first 

 endeavour was to define accurately six points that 

 should severally be good representatives of the six 

 cardinal features in the outline. Those features the 

 limits of which are vague are expressed by italic letters 

 in Fig. 2, and their representative points by the same 

 letters in capitals in Fig. 3. The features are these : — 

 c, the tip of the chin; I, the lower, and it, the upper 

 lip; m, the hollow between the upper lip and the 

 nose ; n, the tip of the nose ; /, the hollow between the 

 nose and the brow. In order to find their respective 

 representative points, proceed as shown in Fig. 2, by 

 drawing (upon tracing paper) a tangent, YY, to both 

 c and /. Then draw a short tangent to n parallel to 

 YY (accidentally omitted in the Fig.). A tangent to 



both c and n intersects the first of these lines at 

 C and the second at N, and determines them. A line 

 drawn from N tangential to / determines F. Thus 

 the fundamental triangle CNF is obtained, in which 

 YCFY is used as the axis of Y, and the length of 

 CF (divided into 100 equal parts, here called " cents ") 

 determines the scale of measurement. In the life- 

 sized portrait of an adult, i cent may be regarded 

 as roughly equivalent to ij mm. or to i/20th of an 

 inch. M, and consequently the triangle CMN, is de- 

 termined by the intersection of one line drawn from 

 C with another from N, both tangent to m. U and L 

 lie at the intersections of tangents drawn in either 

 case, parallel to X and CN respectively. They require 

 less attention than the preceding letters, because u 

 and I are usually small. 



The positions of the six cardinal points may be 

 expressed in either of two ways — (i) as in Fig. 3, by 

 rectangular coordinates, YC\' being the axis in Y, 

 and XCX perpendicular to it, the axis in X. Or 

 (2) as in Fig. 4, by triangulation. Here an addi- 

 tional line, NP, drawn perpendicularly from N to YCY, 

 is convenient. I have compared both of these methods, 

 and found each to have its advantages and disadvan- 

 tages, depending on many variable causes, of which 

 the scale of the portrait is one and the available in- 



