46 
Transactions of the Royal Society of Sotith A frica. 
water is present or not), and appears to be due to increase of molecular 
weight or volume by combination with the environing molecules. Curiously 
enough the nitrate and fluoride form exceptions and scarcely change at 
all with heat ; and cobalt glass and cobalt phosphate would probably be 
pink at low temperatures. Some examples of this change from pink to 
blue may now be discussed. 
a. Cobalt chloride, which in dilute solution shows the ion band at 
X 495, in the solid state has the band shifted a little higher but of the 
same character. If the solution is boiled or treated with a little HCl 
two remarkable very narrow bands develope in the orange and the solution 
becomes violet. These bands are at \\ 607 and 621. Further addition of 
HCl, probably leading to the formation of HC0CI3, causes the solution to 
turn blue, when the original band at about \ 505 disappears and a very 
strong absorption with centre at X 700 in the far red comes up, accompanied 
by another of centre X 661 and a faint one at X 640. In intermediate stages 
all five bands are visible, viz. the three just mentioned and the hair lines at 
XX 607 and 621. When a trace of cobalt salt is dissolved in a large excess 
of cone. HCl the solution is greenish-blue (quite a sensitive test for cobalt), 
and the spectrum shows the bands at X\ 661 and 700 coalesced into one 
band going from X 655 to X 705, i. e. starting at the C line of the sun and 
going half-way to the red end. Probably it is HgCoCl^ which gives this 
spectrum. This band is of course wider if strongly-coloured solutions are 
used and a sixth band near X 530 then appears. 
h. Cobalt bromide does not change appreciably with heat or with con- 
siderable additions of HBr, but if dissolved in HBr (1-38 sp. gr.) and then 
heated the pink solution turns green, and this does not go back at once on 
cooling. The spectrum of the green solution is remarkable, consisting of 
four bands in the red, viz. a pair at XX 637 and 662 (probably due to 
HCoBr.^, since they are narrow and analogous to those attributed to HCoCl^ 
at XX 607 and 621), and a pair of broader strong bands with centres, at 
>X 694 and 735, probably due to HoCoBr^ (c/. XX 661 and 700 in HoCoClJ. 
In deeply-coloured solutions the whole red end from X 680 to X 760 is 
absorbed. The band at X 735 is a most exceptional one, being, I think, the 
only known narrow absorption-band in the far red in any substance ; direct 
sunlight or similar intense light is necessary to enable it to be seen, and a 
violet filter should be used to protect the eye from the glare of the adjacent 
orange light. 
c. Cobalt iodide changes from pink with the utmost difficulty, but a 
yellow-green film was obtained on drying an HI solution of Colg, which 
had a marked band at X 694. This probably corresponds to HC0I3 and to 
X 621 in HC0CI3 and X 662 in HCoBr.,. A very dark brown, practically 
opaque, phase is produced on further drying, which could not be obtained 
thin enough for spectroscopy. 
