15AKUB.] 
PORCELAIN AIR THERMOMETRY. 
191 
obtain the value of the other approximations 1 refer to Tables 41, 42, 
and 43 for the capacities of bulbs and stems. Moreover, an inspection 
of Fig". 37 shows that the 40'" of stem and 150 cra of capillary tube may 
be subdivided into parts, of which the first 5 cm of stem nearest the bulb 
have the temperature T of the bulb, and the 25"" of stem, which project 
out of the furnace, together with the capillary tubes and spaces, have 
the temperature t" of the air. The intermediate 10 cm of stem have a 
varying temperature between T and t", the mean value of which has 
been symbolized by r". Hence the following volumes v, v' v" occur in 
the correctives : 
v = 300™ (Table 43) 
v' =0.0116x10 =0.110 (Table 43) 
v"= f stem, 0.011G x 25 ) (Table 43) 
I copper-cap tube, 100x0.0034 (Table 41) 
\ platinum-cap tube, 50x0.0025 1=1.27 (Table 42) 
I meniscus, 0.4 x 1.22 
[glass cock, 0.2x0.13 \ 
whence it appears that 
!L =0.00039 
v 
—=0.00423. 
v 
With these data in hand it is expedient to return to formula (6), with 
;he object of disposing of the corrective k' =(!-{- xT) 
aiming for a moment that t'=JT, there follows 
Table 44. 
V'l + fif* 
V 1 + 
ar'f 
As- 
T 
k' 
o 
100 
500 
1,000 
1,500 
0. 00045 
0. 00057 
0. 00063 
0. 00067 
lence the error x' is less than 1 : 1000, and it follows reasonably that 
' may be replaced by \T without affecting the result more than T ^ ff 
v" 1-4- fit" 
>er cent. The second corrective K ,/ =(l4-aT) — ~ ' ,, is of greater 
v ^ J v \+at" & 
onsequence, and must be carefully evaluated in each experiment. 
^he practical method consists in calculating an auxiliary table of 
ouble entry for k '+k", in which the arguments are T and t". T in 
uch a table is to vary by successive steps of 100° each, t" in steps of 
0°. From this table, computed once for all, the corrective for any 
alue of Tand i" may easily be taken by graphic or linear interpola- 
(845) 
