504 REPORTS ON THE STATE OF SCIENCH. 
indicator point at the scale (and on the record) when thie bulb is 
surrounded by air. If the indicator point at the scale is not opposite 
the zero marked there, the correspondence is secured by counterpoising 
or more simply by shifting the scale. At the temperature of 18 + 10° 
the zero line would thus be lowered or raised by a weight of about 
20 milligrammes (=rather more than 1 per cent. of chloroform), but 
the correction is not required, since it is made automatically. 
Smoked Cylinder 
Glass Bulb 
Space of 30 litres 
air and chloroform 
vapour at 27 
Variations of temperature during observation would of course raise 
or lower the scale zero, but they are fortunately too trifling to be of 
account in the instrument shown. There is generally a variation of 
not more than + 0'2 to 03°, implying a variation of weight of 0'4 to 
0°6 milligramme, 7.e., 0°0235 to 0°0353 per cent. of chloroform. These 
are negligible corrections in the case of records taken by means of a 
450 ¢.c. bulb with 1 per cent. chloroform at 18° and 760 mm. repre- 
sented by 17 milligrammes of counterpoise and on the record by an 
ordinate of about 3°5 millimetres. 
For finer observations with a delicate balance and a large bulb— 
870 c.c., and 1 per cent. CHCl, represented by 33 milligrammes— 
readings are taken by the null method on the pointer scale with an 
appropriate rider used as counterpoise ; the estimation can then be made 
to an error of + $ or } milligramme, i.e., to an error of between ,}, or 
rks per cent. of CHCl. In this case of course the thermometer and 
barometer readings are taken into reckoning. But for all ordinary 
chloroform determinations temperature and pressure corrections are un- 
necessary. 
The litre weight difference of ether vapour as compared with air 
(2020) is practically one-half that of chloroform (4045), so that the 
chloroform scale 1, 2, 3 per cent. is at the same time an ether scale 
2, 4, 6 per cent. 
