434 Transactions of the South African Philosophical Society. 



Summary. 



1904. 

 Month. 



January . , 

 February 

 March .. 



COBBECTION TO THE MINIMUM TEMPBEATUBE BECOBDED 

 BY NO. 1,470 WHEN IT OCCUBBED AT — 



The Apex of a Sharp Wave. 



Mean 

 Correction. 



-0-76 

 -0-94 

 -0-43 



Maximum 

 Correction. 



o 



2-2 



3-0 



■0-9 



A Period of Steady 

 Temperature. 



Mean 

 Correction. 



-0-06 

 -0-31 

 -0-35 



Maximum 

 Correction. 



o 

 -0-6 

 -0-7 

 -0-6 



The mean correction to No. 1,470 by this method is —0*22°, which corrected 

 for index error of standard becomes —0*32° (c/. pp. 423 and infra). 



The results given in the above table are in entire agreement with 

 those already obtained, and call for no further remark beyond 

 pointing out the very serious nature of these errors, which all 

 tend in the same direction, i.e., to give an erroneous minimum 

 considerably higher than the true one. 



The present investigation leads us to the following conclusions * : — 



1. Minimum spirit thermometers (even the so-called "sensitive") 

 should never be used as ordinary thermometers. 



2. The corrections to the minimum temperature obtained from com- 

 parisons between the readings of the spirit and the readings of a 

 mercurial standard, unless made at an epoch of steady temperature, 

 are unreliable, and may, in extreme cases, be in error to the extent 

 of 2° or more. 



3. The corrections obtained at steady temperatures from com- 

 parisons made as described above, compare favourably with those 

 obtained from comparisons with the lowest points on the Thermo- 

 grams ; but do not, apparently, agree with those obtained in the usual 

 way from comparisons in water (cf. pp. 423, 432 et supra). 



4. The absolute minima obtained with spherical bulb ther- 

 mometers are frequently in error on account of sluggishness, the 

 errors in extreme cases amounting to as much as +3°. 



5. The mean of the absolute minima obtained with spherical bulb 

 thermometers is invariably too high, as all errors due to sluggish- 

 ness tend in the same direction. 



* This is only intended to apply to the types of thermometers employed in this 

 discussion ; it is quite possible that, at other stations, minimum thermometers of 

 greater sensitiveness are in use. It should, however, be remarked that three 

 spherical bulb thermometers made by different makers in the years 1859, 1878, 

 and 1894 respectively give practically the same value of A. 



