ME. V. CEOOKES ON THE ATOMIC WEIGHT OE THALLIUM. 
283 
suspension-wires of the pans could be hung. This required that the pans should not 
differ from each other by a quantity greater than one thousandth of a grain. 
In heavy weighings it is found convenient to remove one of the pans ; but as the case 
is one of determining a weight and its increase after certain operations, the removal of 
the pan does not affect the result, provided the weight of the pan is accurately ascertained 
and this weight allowed for, the apparatus weighed appearing lighter to an amount 
equal to this weight. Always when weighing different metals, or glass, or some chemical 
against metal, it is necessary to correct for the weight of air displaced, reference being 
at the same time made to the temperature and air-pressure ; for assuming that there are 
to be weighed 7000 grains of bronze against 7000 grains of platinum, there will arise 
an error of nearly Offi of a grain unless this precaution be attended to, for 7000 grains 
of bronze displace roughly 1 grain of air, while 7000 grains of platinum displace only 
0'4 grain. 
At each weighing at diminished air-pressure care must be taken to allow the balance 
to remain at rest for at least half an hour, and preferably for several hours, in order to 
allow the temperature to become uniform after the alteration caused by the exhaustion. 
The weighings were always repeated a second time after every thing had been allowed 
to remain at rest for one hour ; and when the final weighing was made, the case had 
been unopened for six hours, the adjustment being made by slightly altering the density 
of the enclosed air. 
One of the greatest difficulties occurred in endeavouring to illuminate the scale and 
pointer of the balance without heating sufficiently to introduce a cause of error. The 
concentrated rays of a lamp were found to be unsuited in several ways. The use of a 
small vacuum-tube suspended inside the iron case was finally decided upon, sufficient 
light being obtained with two Grove’s cells actuating a small induction-coil placed some 
distance from the apparatus, the electricity being conveyed by fine conducting-wires of 
good copper, carefully insulated*. These wires pass into the case through grooves filed 
in the flanges and well protected with india-rubber bands, and in no way interfere with 
the obtaining of a vacuum. 
To prevent parallax the scale and pointer are viewed through a small telescope having 
a vertical wire in the focus of the eyepiece. The observer is therefore able to be situ- 
ated some eight or ten feet from the balance during accurate observations, thus reducing 
to a minimum the disturbance due to the temperature of the body. It is inexpedient 
to estimate the value of a division on the ivory scale over which the pointer of the 
balance travels, as its value varies with the length of arc of vibration, with the weight 
in the pans, and slightly with the temperature. It is also evident that Gauss’s method 
of weighing in alternate pans is inapplicable when weighing in a rare atmosphere, 
owing to the number of times the case would have to be opened, and the consequent 
liability to other sources of error. Borda’s method, as described by Peclet in his 
* At high rarefactions this method of illumination fails, owing to the induced current passing between the 
wires outside the vacuum-tube. ' 
