590 EEPORT— 1888. 



mechanism to be attached to each instrument is very simple. A thermometer 

 fitted with it gives satisfactory results. 



An arrangement whereby an indefinite number of instruments may be con- 

 trolled by one light cable of three wires is proposed. It has not, however, been 

 practically tested. 



9. On the Mechanical Conditions of a Swarm of Meteorites, and on Theories 

 of Cosmogony. By Professor G. H. Daewin, F.B.S. 



10. 0)1 some accurate Charts of Kew Corrections for Mercury Thermometers. 



52/"W. K Shaw, M.A. 



An English thermometer (Hicks, No. 79915) graduated in the stem to 0°-2 C. 

 was compared at Kew in the ordinary way in December 1880 to 0'1° 0. at each 

 fifth degree between 0° and 35°. In January 1882, through the kindness of Mr. 

 G. M. Whipple, Superintendent of the Kew Observatory, it was compared at each 

 degree and the readings were estimated to the hundredth of a degree. A similar 

 comparison was carried out in the same way in April 1888. The errors from true 

 (Kew standard) temperature are plotted and the points joined to facilitate inter- 

 polation. For the two sets of more accurate comparisons the curves are smoothed 

 within the limits of possible errors of observation. These two curves are drawn 

 on the same sheet with the plotting of the original comparison and the juxta- 

 position makes it easy to compare the curves. During the interval from January 

 1882 to April 1888 tlae zero had risen 01° C, and, on the usual assumption, the 

 two corresponding curves should run parallel to each otlier at a distance repre- 

 senting that difference of temperature. The general similarity of the two curves 

 is very fairly exhibited in the diagram, and affords evidence of accuracy and trust- 

 worthiness of the comparisons upon which Mr. Whipple and his staff are to be 

 congratulated. A slight divergence from parallelism is, however, clear, as the 

 curves get gradually wider apart as the temperature rises, the difference at 30^ 

 being 0'17° C, whereas at 2 it is 0'1° C. The original rougher comparison 

 epitomises the more accurate ones in a very striking manner. 



Similar results follow from charts for two other English thermometers by 

 Hicks and for two German thermometers by Geissler. But the divergence from 

 parallelism is not shown in the charts of two thermometers by Negretti, nor of two 

 by Casella, the curves in these four cases running very nearly parallel, although 

 there were changes of zero of 0'35° F. and 0-2° F. respectively. In the case of the 

 German thermometers there was no appreciable change of zero in the interval. It 

 would appear that the divergence from parallelism is due to the nature of the 

 glass employed in the construction of the instruments. 



11. On an Apparattis for determining Temperatiire hy the Variation of 

 Electrical Resistance. By W. N. Shaw, M.A. 



This apparatus is designed to measure to a high degree of accuracy the mean 

 temperature of a large burette in a water-bath. A Wheatstone quadrilateral 

 is formed by joining the ends of a platinum wire about 30 inches long (resistance 

 about 11 ohms) to those of a platinum-silver wire BCDA. The platinum 

 wire is bare and the alloy is silk covered, '007 inch in diameter ; its resistance i& 

 13'5 ohms per metre. At A, and B, D respectively battery and galvanometer 

 wires are soldered ; the lengths of the wires are so arranged that for some tempera- 

 ture near 15° C. there is a perfect balance indicated in the bridge. The wires are 

 laid in parallel lengths and disposed upon a strip of pure india-rubber, about three- 

 quarters of an inch wide, in the manner indicated in the figure. The india-rubber 

 is painted with a solution of india-rubber in benzene and a second strip, similarly 

 painted, is laid down upon it and the wires are thus completely enclosed ; the 

 strip contains 18 inches at each end beyond the junctions of leading wires, so that 



