94 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 54 



iodine actually employed, the other from the quantity of iodide of silver 

 collected. From the first set we have of iodine for 100 parts of silver: 



117.5390 

 117.5380 

 117.5318 

 117.5430 

 117.5420 

 117.5300 



Mean, 117.5373, ± .0015 



From the weight of silver iodide actually collected the following figures 

 are given for the ratio Ag : I. The third experiment in the foregoing 

 column has no equivalent here : 



117.529 

 117.531 

 117.539 

 117.538 

 117.530 



Mean, 117.5334, ± .0014 



These determinations, by Marignac and Stas, are remarkably con- 

 cordant, and yet, as shown by later investigations, they are affected by 

 constant errors. Silver iodide, precipitated from nitrate solutions, oc- 

 cludes silver nitrate, a fact which must be taken into account in two 

 of the preceding series. The concordance between the second and third 

 series of Stas, however, remains unexplained, if we suppose them to be 

 in error also. That the errors in four sets of determinations, by two 

 observers and four methods, should be so exactly alike in direction and 

 magnitude, is difficult to understand. 



With Ag= 107.93, the Marignac-Stas determinations of this ratio 

 make 1 = 126.85. This value was accepted for many years, until Laden- 

 burg, in 1902, shojved that it was about one-tenth of a unit too low. 

 Ladenburg ^ depended upon the ratio Ag : AgCl to establish this con- 

 clusion, but he also gave one measurement of the ratio now under con- 

 sideration, as follows: 50.3147 grammes Ag gave 109.4608 Agl, whence 

 the ratio Ag: Agl = 217.552; a figure higher than those in the fore- 

 going tables, but not strikingly so. 



Soon after tlie publication of Ladenburg's memoir, Scott ^ announced 

 two syntheses of silver iodide, as follows, with weights corrected to a 

 vacuum : 



1 Ber. Deutsch. chem. Ges., 35, 2275. 19fi2. 

 M'roc. Chem. Soc, 18, 112. 1902. 



