November 14, 1895] 



NATURE 



45 



The separate determinations were entirely independent, and 

 taken by three different observers ; the coils were so changed 

 that the b.-w. readings altered from - 7*263 to + 12784, 

 while the sum of the corrections varies from + 0*131 to + 0*043, 

 yet the greatest departure from the mean = 0*003. 



Thermometer Kj in Steam, 



Date and 

 observer. 



Oct. 2, 1895 



C.T.H. 



E.H.G. 



Coils and 

 correction. 



12.37 



12.42 



C.D.F.FI 

 = 360 



357300 

 357-296 



The barometer reading at 12.37 corrected for temperature, 

 and for ^ to sea-level, lat. 45° = 750*12 m.m., and at 12.42 = 

 75006 m.m. (This difference of '06 m.m. corresponds to a 

 decrease of o 002 in R . ) 



Hence mean pressure = 750 09 m.m., and tenrtperatu e of 

 steam at this pressure = 99" '634 C. 



.-. 347.'J98 - 257*827 ^ ^,^„ change in R per 1° C. over this 

 99*634 " 



range. 



Now ^M =.\ - 5 - " ^°°. We may assume 8 for this wire 

 5/ 10,000 



as approximately 1*50. 

 Hence 



^' = 0*985, .-. ^^ at 100° = -9982 X -985 = *983, 



Hence 



.*. 5R1 for 0*366° C. = 0*360. 



Ri = 357-298 + *36o = 357-658. 



Thermometer K^ tn Suiphur-vapour. 



Here, again, changes in coils and b.-w. readings do not 

 aijpreciably affect the results. It is interesting to notice that 

 the change in box temperature between 1.58 and 2.20 \t.xn. 

 almost exactly accounts for the difference of 0*021 in the b.-w. 

 readings by the different observers. As the apparatus had but 

 just been installed, we had not got the regulator properly under 

 control ; thus the temperature changes were greater than would 

 usually be the case. 



Barometer, corrected as before = 750*30 m.m. 



NO. 1359, VOL. 53] 



Now, b. p. of sulphur 



= 444-53 + (/ - 760) X *o82 = 443-73. 



and 



^,^^677-775-257-827^ .66. 

 357 658 - 257-827 



Hence, Eq. (d), 



443-73 - 420*66 = 8(4*4372 - 4*437), 



.*. 8 = 1*512. 



Thermometer K, is now completely standardised. 

 Constants 



Ri = 357-658 ^=1-3872 



R. = 257*82^ 



FI = 99*831 8= 1*512. 



The most simple manner of obtaining the value of t for any 

 given value of// is to proceed as follows. Construct, by means 

 of Eq. (d), a table giving corresponding values of // and d for 

 regular increases in t or//, assuming the value of 8 as i -500. (For 

 convenience of those using these thermometers, I give such a 

 table, as an appendix, for values of// up to 1000.) Plot the 

 numbers thus obtained on a large scale with // as abscissa and d 

 as ordinate. Having experimentally found // in a certain case 

 with a thermometer whose value of 8 is 8\ ascertain from the 



chart the corresponding value of d, then t — pt + • x d, 



1*500 

 and thus the same chart can be used for different thermometers. 

 The above example will suffice both to illustrate the general 

 method and the standardisation of the Kew thermometers. 



IV. Concluding Remarks. 



I understand that the Kew Committee had two objects in 

 view when they sanctioned the acquisition of this apparatus and 

 undertook the task of directing a course of observations; 



(i) To submit the methods and principles of platinum ther- 

 mometry to an exhaustive trial, and especially to ascertain how 

 far the apparatus would stand the test of time and use. Such a 

 series of observations can only be undertaken by a department 

 similar to that at Kew, where records are properly kept, and 

 where the continuation of the experiments is not dependent on 

 the life or inclination of individual observers. 



(2) To establish some recognised system of standardisation 

 for instruments intended for the measurement of high tempera- 

 tures. 



With regard to the latter object, I would venture to add a few 

 remarks. If high-temperature mercury thermometers (such as 

 those of Niehls, of Berlin) are sent for comparison, it must be 

 remembered that the ' readings of these instruments are greatly 

 influenced by the stem temperature, especially when the range 

 is large, and it is impossible, under the conditions usually prevalent 

 in high temperature measurements, to secure complete immersion 

 of the stem. The observers at Kew will be able to state the length 

 of the portion actually immersed, &c., and those who afterw-ards 

 use such thermometers mUst endeavour, if they wish for accurate 

 results, to reproduce the conditions as nearly as possible. 

 The experience of MesSifst Heycbck and Neville in their earlier 

 work (when they uied • for theit experiments mercury thermo- 

 meters standardised by platinum ones) ^ shows that it is possible 

 to reproduce the original conditions -with sufficient accuracy. 



Again, it is useless to standardise glass thermometers unless 

 previous experience has shown that they are not subject to t he- 

 zero rise usually characteristic of such instruments after exposure 

 to high temperatures. 



Another matter, to which I trust the attention of Dr. Chrcc 

 will sometime be directed, is the suitability of platinum 

 standards for the calibration of mercury thermometers at ordintry 

 temperatures. The greatest value of d over the range 0° to io<j 

 {i.e. near 50° C.)is with these standards less than o''*4C. ; n).v 

 an error of i per cent, in 8 (and I do not believe that any su h 

 error is probable, pr I may say possible) would mean an error of 

 but o°-oo4 in /at 50° C., and less at other temperatures. The ro.ul- 

 mgs are independent of changes in internal or external pres>u c. 

 of position or of stem immersion, and are directly expressed \\\ 

 terms of the air thermometer. It was with a view to such com- 

 parisons that I designed Kg, which on account of the large value 

 of FI would cause an error of *oo3 in the readings to affect the 

 resulting value of / by only o°*ooi C, 



1 Ckem. Soc. fcur.t., July 1890. 



