90 REPORTS ON THE STATE OF SCIENCE. 
Comparison of two Dines’ Thermographs.' 
Temperature | Temperature 
Height = ——|| Height 
No. 90 | No. 94 No. 90 No. 94 
Km. | Km. 
1 —1 — 38 eel ice | —56 —54 
2 —7 | -—9 (arnt os —56 —54 
3 -13 —15 | 1d —55 —54 
4 —19 —22 Lp ied —55 —54 
5 — 24 / —26 | 16 —5b5 —56 | 
6 — 30 —32 lol ee —55 — 56 | 
7 —34 -- 37 18 —56 —57 
8 — 40 | — 43 he eis —57 —57 
9 _ —46 | — 50 i 20 —58 —59 
10 -- 52 — 54 ee ee a —59 —60 
11 —56 — 54 | 22 | — 60 — 60 
A. de Quervain in 1906 estimated from the results of his own and 
Hergesell’s experiments the probable maximum and mean errors of the 
various thermometers: due to lag.? His results may be tabulated as 
follows :— 
| Maximum error for fall | Probable mean error 
Assmann Copper-Invar 
Bosch New Bimetallic : : 
Kusnetzow (ring-shape bimetallic) . 
Instrument of 4° C. per min. with | under ordinary work- 
ordinary ventilation ing conditions 
Hergesell-Bosch Tube : ; 0:95° C 00°C 
Teisserenc De Bort New Bimetallic. 10 05 
x 4 Old 5) 2°5 0-7 
Bourdon Tube . ; : 2°5 _ 
3°8 14 
2°4 
2°2 
In continuation of the work of Hergesell and Maurer he examined 
(1) the small Bourdon tube thermograph of Teisserenc de Bort : German- 
silver tube slightly curved and filled with alcohol; (2) the bimetallic 
lamellar thermograph of Teisserenc de Bort: strips of brass and steel 
soldered together and bent into a circle; (3) the tube thermograph of 
Hergesell : German-silver tubes on supports of nickel-steel ; (4) the bi- 
metallic thermograph of Assmann: copper and nickel-steel; (5) the 
Kusnetzow bimetallic thermograph : brass and nickel-steel bent into an 
S curve. The first three had often been used simultaneously. 
He concluded that (1) and (2) were in good agreement, (3) generally 
marked 0°6 C. less than (2), and indicated lower amplitudes at sudden 
changes, though on the whole more sensitive than (2). 
He concluded that the mean error for the pressure records was +2 mm. 
and for the temperatures +2°C., so that the temperature gradients for 
1000 m. intervals are reliable to a few tenths of a degree C. 
A slight error is introduced into the bimetallic thermometer calibrations 
through change of zero by straining of the soldered joint. The Hergesell- 
' Weekly Weather Report, February 28-March 6, 1909, p. 76. 
* Beitr. z. Physik der Fr. Atmos. i. p. 163, 1906; Zeitschr. SF. Instrumentenk., 27, 
pp. 127, 128; April, 1907. 
