130 REPORT— 1890. 
different wires; unless working with long lengths of thick wire the 
weight of the wire is small. The weight of the water displaced cannot 
be determined within ‘5 to 1 milligramme, and that only with care: this 
error in ‘5 of a gramme means only an accuracy of 1 in 500. The values 
given in my table are probably correct to 1 in 1,500 or 1 in 2,000, as the 
weight of water displaced was in all cases over 2 grammes. Results, 
therefore, for resistances of wires of equal weight are the most trust- 
worthy, and, I think, also the most satisfactory if used to express the 
resistance of a material and not of any given wire. 
Wires X. (1) and X. (2) are of the same copper, but drawn down 
separately: X. (1) was beginning to fray, and another specimen of the 
same copper drawn down still further had on this account to be re- 
jected; this has affected the resistance value expressed in both ways. 
hus :— 
> MES) BE j «3. oe : ee A 8 
2 (ES : . 1569 : ct leone 
but much more so when expressed for equal volumes. In both the 
copper is of the same quality. 
It will be noticed that with increase of specific gravity there is a 
decrease of resistance, even when the results are expressed for wires of 
equal weight. The resistance diminishes, therefore, more rapidly than 
the density increases. Wires of the same quality may, in consequence 
of a difference in drawing, have a different density, and so the resuits 
expressed in terms of equal volume will differ considerably, while those 
for equal weight are the same, or approximately so. 
The values obtained for IX., X., and XI. are so nearly identical that 
it is not unfair to conclude that they are samples of pure copper; their 
value is identical with that obtained by Matthiessen at, I believe, the 
same temperature. The greater difference obtained at 0° C. between 
Matthiessen’s value and samples of copper tested now at that tempera- 
ture is probably due to the fact that Matthiessen’s value was not 
determined at 0°, but reduced in value for that temperature from observa- 
tions, as stated above, at about 20° C. 
The higher conductivity or less resistance for the two samples given 
in the table is due, not to increased purity in the preparation of the 
copper, but to the difference in the process of preparation, whereby .a 
sample of greater density is obtained than results from the working up 
of small quantities of copper in the laboratory. 
A sample of copper has been prepared by chemical means with the 
help of my friend Mr. Skinner, but has not yet been measured. 
APPENDIX IV, 
A Comparison of a Platinum Thermometer with some Merewry Thermometers 
at Low Temperatures. By EH. H. Grirrirus, M.A., Sidney College, 
Cambridge. 
The following communication describes the mode of constructing 
an air-tight platinum thermometer for use at low temperatures. The 
thermometer was graduated by means of the freezing and boiling points 
