81 
of Edinburgh, Session 1866-67. 
difference between the squares of the interval in the one case and 
in the other will be proportional to the change in the specific 
gravity of the water, which again, when the amount of impurity is 
minute, may be held as being proportional to the amount of that 
impurity. 
The determination, by this method, of the quantity of heavy 
impurity in potable water may be freed from all dependence on a 
knowledge of the co-efficients of the expansibility of water, by com- 
paring the observations made on distilled water with those made 
on water containing a known quantity of lime or other saline 
matter. And this, indeed, is the only useful method of procedure ; 
for in the apparatus exhibited there are many glass balls, and, to 
distinguish these from each other, they are made of glass of various 
colours ; and it is found that the means of the temperatures at which 
these pass the thermometer are different for the different colours. 
These means are also influenced by the conformation of the balls ; 
thus those which are thin, and which, consequently, have heavy 
drops, are more acted on by the expansion of the internal air, than 
those are which have thick sides, and, consequently, small drops. 
By using a glass ball adjusted to float in pure water just at its 
maximum relative density, we obtain the greatest delicacy in the 
indications. I found that such a ball, when placed in the water 
supplied to the New Town of Edinburgh, gave an interval of 17 2 
degrees on Fahrenheit’s thermometer. And the same method may 
be extended to the case of balls which float on water of much 
higher temperatures, only it then becomes necessary to use a ther- 
mometer of proportionally greater delicacy in its graduations. And 
it may be remarked that, for the purpose of securing certainty in 
the indications, it is expedient to inclose the jar containing the 
water to be experimented on in a large vessel also containing 
water, and to agitate the water contained in it so as to keep the 
fluid in the internal vessel as nearly as possible at the same tem- 
perature from top to bottom. By this arrangement we almost en- 
tirely avoid the formation of internal currents, which would tend to 
give uncertainty to the observations. 
VOL. VI. 
L 
