RICHARDS AND I'ARKKR. — ATOMIC WEIGHT OF MAGNESIUM. 69 



stnu'ted for the purpose, and consisted of two rectangular glass cells, with 

 a mirror enclosed in a dark case, so arranged that the column of liquid 

 contained iu the lower part of the cells could be viewed horizontally with- 

 out disturbance from surface reflections. A dark scieen was placed at 

 the further end of the cells, and the whole so arranged tliat light could 

 come to the eye only by reflection from solid particles which might be 

 suspended in the oolunni of liquid inspected. If the liquid was perfectly 

 clear, the field of vision remained black, but an extremely small amount 

 of precipitate produced a very marked change, and the intensity of opales- 

 cence was approximately proportional to the amount of precipitate. It 

 was found perfectly easy and certain, by this method, to distinguish the 

 difference between .002 and .003 of a milligram of argentic chloride or 

 between .004 and .005 of a milligram, and larger amounts in proportion 

 Tliis instrument gave such satisfaction in this research tiiat the method 

 will be worked out for various other reactions, and published later. 



The method of using this apparatus was as follows. Twenty-five 

 cubic centimeters of the clear supernatant liquid from the flask contain- 

 ing the well shaken argentic chloride and magnesic nitrate were placed 

 in eich cell, five cubic centiijieters of a very dilute solution of argentic 

 nitrate being added to one, and five cubic centimeters of a correspond- 

 ingly dilute solution of ammonic chloride to the other. The silver 

 solution contained one milligram of silver to the cubic centimeter. An 

 unequal depth of cloudiness indicated an excess of either silver or 

 chlorine iu the original solution, and accordingly the amount necessary 

 for neutralization was run into the large flask containing precipitate and 

 solution from a burette. The solution was again allowed to stand in the 

 dark with occasional shaking, and after the precipitate had entirely sub- 

 sided was again tested as-before, and this cycle of operations was repeated 

 until the opalescences matched one another.* It will be observed that, 

 if water is added to the cell giving the more dense opalescence until the 

 effect becomes equal on both sides, the amount of dilution will give a 

 means of ascertaining the amount of precipitate in each cell. The appro- 

 priate corrections were then applied to the amount of silver taken. Due 

 allowance was made for the slightly diminishing volume of the solution 

 in the flask. The addition of one tenth of a milligram of silver to a litre 

 of solution produced a distinct change in the depth of color observed. 

 After the matching was completed, repeated trials were made with fresh 



* For details of this method, see Stas, Mem. Acad. Belg., XLIII. Part II., and 

 Richards, These Proceedings, XXIX. 86; XXX. -385. 



