118 Chronicles of @cience. | Jan., 
recognized by the lines at 105 and 109 and by the brightness near 
85 and 87; the bromide compound .is detected by the brightness at 
85, 884, and 92; and iodide of copper by the brilliancy at 96, 99, 
and 1023. ‘ 
By this method, and without further trouble, + per cent. chlorine, 
$ per cent. bromine, and 1 per cent. of iodine are easily recognized, 
and a practised observer may detect much smaller quantities. 
When the haloids are mixed with each other in very small pro- 
portions, it is better to precipitate them first by a silver salt. Mix the 
dried precipitate intimately with twice its weight of oxide of copper, 
and employ this mixture in the hydrogen before described. By this 
method as little as yoth per cent. of chlorine, ;th per cent. of bro- 
mine, and }th per cent. of iodine in the silver precipitate can be 
recognized, 
The spectra of the haloid salts appear consecutively, that of the. 
chloride first, then that of the bromide, and lastly that of the iodide 
of copper. Their appearance in this order depends on the different 
volatility of these salts. Chloride of copper volatilizes considerably 
below a red heat, the bromide somewhere near redness, and the 
iodide at a low red heat. The slower the volatilization is conducted, 
the more certain are the results of the analysis. 
When only traces of iodine and bromine compounds are present 
in a large excess of a chlorine compound, about the tenth of a 
gramme of nitrate of silver should be added to the solution. ‘The 
greater part of the iodine and bromine will be found in the precipi- 
tate, which may be tested as before described. 
The following results will serve as an illustration of the accuracy 
and precision of the process. ‘To a pound of common salt which 
contained no bromine five milligrammes of bromide of sodium were 
added, and to the solution one decigramme of nitrate of silver. The 
precipitate was tested in the way just described, and after the spec- 
trum of chloride of copper had been observed for some time, the 
spectrum of bromide of copper was distinctly visible for five minutes. 
A further addition of nitrate of silver to the solution gave a precipi- 
tate which showed the spectrum for six minutes. 
Similar experiments made with iodine compounds gave equally 
conclusive results, and proved that a ten-millionth part of iodine or 
bromine may be detected in chloride of sodium. 
The residue of six and a half pounds of sea water taken off 
Heligoland showed the spectrum of the bromide for seven minutes. 
Iodine could not be recognized, probably because the quantity of 
water was too small. 
A small quantity of water from the Dead Sea showed a large pro- 
portion of bromine, but no iodine. The mother-liquor from some salt 
works showed much bromine, but no iodine. 
Some valuable contributions to our knowledge of spectrum ana- 
