DIVISIBILITY OF GOLD AND OTHER METALS. 79 



films were superposed, one upon the other, the color was very per- 

 ceptibly darkened; even when subjected to a magnifying power of 

 1,000 diameters the films retain their continuous character, though 

 the brilliance of the green color is of course diffused. The dimensions 

 of the waves of light, when decomposed by the prism, have been care- 

 fully measured. There are 47,000 green waves in the space of an inch ; 

 dividing the estimated tliickness of our gold film by this number, we 

 find that the thickness of the film is less than gL part of a single undu- 

 lation of green light. 



In the course of an examination made by the writer upon " The 

 Practicability of assaying Metals used in Coinage, by means of Spec- 

 trum Analysis, made in and for the Assay Department of the United 

 States Mint at Philadelphia," * it was noticed that a large number of 

 very powerful electric flashes might be passed between two slips of 

 metal without any apparent loss, and an important query suggested 

 itself, viz., whether the amount of metal vaporized by each spark was 

 not too infinitesimal to be determined. In order to ascertain this 

 p<nnt the following experiments were tried : Having weighed small 

 electrodes averaging 18 milligrammes each with the greatest possible 

 accuracy upon the gold assay-balance of the mint, which is sensitiA-e 

 to ^ of a milligramme, or y^Vr ^^ ^ grain troy, and having arranged 

 a spark- register, it was found that 1,000 sparks might be passed be- 

 tween these poles, each sjaark showing the spectrum of the metal dis- 

 tinctly, and yet the loss in weight was too small to be made the base 

 of calculation. Thus a gold pole lost in weight, after passing 1,000 

 sparks, yoVo ^^ ^ grain ; this gives for each spark yoTrJinro of a grain 

 of gold, producing a bright spectrum. 



The number was then increased to 3,000 sparks as the test. The 

 loss of weight depends, of course, upon the electric volume, and in the 

 experiments tabulated an endeavor was made to keep the latter con- 

 stant. The tables (marked A and B) show that the loss in weight is 

 marvelously small, averaging less than -^ of a milligramme of gold 

 for 3,000 sparks. To give the amount for each spark, this must be 

 divided by the number of sparks ; thus, in round numbers, an elec- 

 trode loses YoVo of ^ grain after ])assing 3,000 sparks; or, for 1,000 

 sparks, y^Vo of a grain ; or, for each spark, yo-qwoo" of a grain. The 

 experiments made by M. Caj^pel to determine the minimum amount 

 of each element that would show a spectrum have been published in 

 tabular form. His method was to volatilize " solutions of the metallic 

 salts between the poles of a small induction-coil in Mitscherlich's glass 

 tubes with platinum wicks. A series of solutions, each one-half the 

 strength of the preceding one, was prepared from a number of metal- 

 lic chlorides. The spectrum in connection with the positive pole was 

 continually observed, while increasingly-concentrated solutions were 



' Published in the " Proceedings of the American Philosophical Society," vol. xiv., 

 p. 162. 



