908 
MR. W. CROOKES ON RADIANT MATTER SPECTROSCOPY: 
brought down a heavy crystalline oxalate. This was filtered off, and called oxalate A. 
The filtrate, again heated to boiling, was precipitated in exactly the same way with a 
further quantity of amnionic oxalate till the hot liquid became opalescent. On cooling 
and stirring, a farther quantity of oxalate came down. The filtrations and 
precipitations were repeated until no more precipitate could be obtained. Usually I 
could get twelve or thirteen fractionations in this manner ; towards the end the 
solution did not get milky, and it had to stand sometimes twenty-four hours before 
much oxalate came down. 
52. The fractions first precipitated by oxalic acid gave very strong absorption bands 
when the concentrated solutions of the oxides were examined by transmitted light. 
The fractions last precipitated showed the absorption bands only faintly. 
53. These operations gave me oxalates from A to L. These, ignited, with free 
access of air, were then each dissolved in nitric acid, and again separately fractionated 
as oxalates. The result was about 150 precipitates, ranging from A x Ao . > . A 12 , 
hj fij ... to Lg ... h^* 
T1 lese, after ignition, were separated into five lots according to order of colour, and 
the fractionation of each of the five lots repeated as already described; the series of 
operations now closely resembling those of Pattinson’s process for desilvering lead. 
This gave me about sixty lots. This time the hydrogen equivalent of the metal of 
each lot was taken by converting the oxalate into sulphate and estimating the 
sulphuric acid, assuming M 2 0 to be the type of oxide (40, note 1). The result was a 
series of earths having hydrogen equivalents (M) ranging from about 48 to 33. The 
earths were now sorted into high, low, and intermediate, those giving intermediate II 
equivalents being re-fractionated with repeated H equivalent estimation, the highest 
and lowest being each time separated and added to the former high and low lots. 
54. The ultimate result of about five hundred fractional precipitations gave me a 
mixture of earths having an H equivalent M = 48, and showing a strong absorption 
spectrum (56); a mixture having an H equivalent M = 33, having no absorption 
spectrum (65); and intermediate earths. 
In the radiant matter tube all these fractions gave the citron-band spectrum well, 
but that of the earth of lowest equivalent was much the brightest, and that of the 
highest equivalent the least intense, 
55. Three methods are available for the partial separation of these earths and for 
the complete purification of any one of them. The formic acid process (56, 57) is best 
for separating terbia, as terbic formate is difficultly soluble in water, the other 
formates being easily soluble. 
Fractional precipitation with oxalic acid (63, 64, 65) separates first erbia, holmia, 
and thulia, then terbia, and lastly yttria. This is the only method which is applicable 
for the separation of small quantities of terbia from yttria. 
Fusing the nitrates (60, 68, 69) separates ytterbia, erbia, holmia, and thulia from 
yttria. It is not so applicable when terbia is present, and is most useful in purifying 
