﻿682 On the Absolute Expansion of Mercury. 



three times the linear in each case. I became aware of this 

 coincidence some two years ago when the first results with 

 the silica -weight thermometer were obtained in my private 

 laboratory, but w T e were unable to detect any source of error 

 in our experiments sufficient to account for the discrepancy, 

 and we were not in a position to discuss it at the time when 

 our paper was published. I am at present inclined to the 

 opinion that the explanation must be sought in the inequality 

 of axial and rad'al expansion of a drawn tube due to intrinsic 

 strain. It is difficult to see how this effect, if it exists, could 

 be satisfactorily detected except by reference to the absolute 

 expansion of mercury. The intrinsic strain in a hard glass 

 tube appears to be of the same order as in a silica tube when 

 tested by polarized light, and would probably produce the 

 same absolute effect on the expansion. On this view it is 

 necessary to suppose that the anomaly giving rise to a 

 minimum length for a silica rod between —50° and —80° (\ 

 affects the radial expansion at a somewhat higher temperature, 

 giving a minimum in the neighbourhood of 0° 0. It is at 

 least noteworthy that, if our results for the absolute expan- 

 sion of mercury at 184° C. are assumed, the cubical co- 

 efficient of silica is already nearly three times the linear, 

 when the latter has nearly reached the constant value which, 

 according to my experiments*, it retains over the range 

 300° C. to 1400°" C. But for the anomaly in the expansion 

 of silica at low temperatures, and the impossibility of an- 

 nealing it satisfactorily owing to its rapid devitrification at 

 temperatures above 900° C, a silica bulb, employed according 

 to Chappuis's method, would afford an almost ideal method 

 of determining the expansion of a liquid. The measurement 

 of the absolute expansion of mercury by the hydrostatic 

 method is a difficult experiment, and may possibly involve 

 sources of error which are not taken into account in the 

 accepted theory. These are being further investigated, and 

 the observations with the silica weight thermometer are being- 

 extended to higher and lower temperatures. It may be 

 objected that such investigations should not be undertaken 

 by amateurs in their occasional intervals of leisure, but should 

 be confined to National Physical Laboratories, where the ex- 

 perimentalist may devote his whole time and attention to one 

 research. In our case it may be urged in extenuation that the 

 method was designed, and the chief part of the apparatus con- 

 structed, before the National Physical Laboratory existed. The 

 work has been carried through under considerable difficulty, 

 and is doubtless not so perfect as it should be. The apparatus 

 was designed for working up to 300° C, and is perhaps not 

 * Shenstone, Tree. Boy. Inst. IDOL Not quoted by Scheel & Heuse. 



