576 
The above probable errors include the periodical summer depression and 
winter ascent ; doubtless due, as Drspretz first remarked, to the effect of tem- 
perature. Thermometer (DM) appears to have been rather irregular in its 
indications ; (AE) and B, which were examined at the same time with it, gave 
MR EDMUND J. MILLS’S 
far too erratic numbers for any attempt at reduction. 
The following experiments on thermometers 20, 50, 101, and 454 were con- 
ducted by myself. All the instruments had cylindrical bulbs; they were all 
vacuous, and had just been heated to 100°, with the exception of 20, which had 
just been opened. The unit of x is 30 days. 
The equations are— 
Yoo = '214(:7565)* + 260(:008)’, 
Yoo = '159('8300)* + 151(-080)", 
Vo = '144(-88116)* + 046 (4534), 
Yas4= '0545(-8840)* + °116(-112)’. 
Therm. | Therm. 
x | 20 20 
Found. | Cale. 
| 
0:00) 2 jae 
0:30 | 209 | -247 || 0:33 
0:70 | 180 | 185 || 0-93 
173 | 090 | 131 || 167 
2:67 | 139. | 103. || 3:33 
2:87 | 090 | 098 || 6-67 
517 | 090 | -050 || 11-47 
6:87 | 027 | -030 | 
10-47 008 | -011 
Probable error 
of one com-| | 
parison, . |+°020| 
Observations with thermometer 20, before opening, had furnished the following 
equation,— 
TABLE IV. 
Therm, | Therm. Therm. 
50 50 ai 101 
Found. | Cale. Found. 
— — 000; — 
178 | ‘214 1:00 | 134 
146 | 149 200) 106 
“147 | 118 4:00 | :082 
089 | :086 5:00 | :076 
‘047 | 046 | 6:00} :068 
000 | 019 || 17-00.| -073 
17:03 | :009 
(21°08 | ‘021 
015 | 
Therm. 
101 
Cale. 
‘006 
y = '212('73738)" + °256(:0430)", 
the unit of # being 10 days. 
given in Table V. 
A comparison between theory and experiment is 
Therm. 
4 
Found. 
170 
‘067 
‘048 
030 
‘030 
026 
‘000 
Therm. 
454 
Cale. 
‘005 
