Dixon and Atkins — On Osmotic. Pressure in Plants, ^c. 279 



prevent short oireuiling and the oliemiual action of the solutions on the 

 junctions. 



The sohition whose freezing-point was to be determined and the distilled 

 water which was to act as the standard were put into two small test-tubes 

 about 10 cm. long and 1 cm. iu diameter. These were supported in a large 

 cork bung which fitted into the neck of a cylindrical glass vessel. The latter 

 could be immersed up to tlie neck in a larger vessel which contained a 

 freezing-mixture of ice and salt. 



The galvanometer emjDloyed is an Ayrton-Mather instrument manufac- 

 (ured by the Cambridge Scientific Instrument Co. Its resistance is 20'7 w. 

 The deflection at one metre for one micro-volt is equal to 10 mm., and for 

 one micro-ampere 206 mm. The translucent screen was set at 94 cm. 

 distance from the mirror of the galvanometer, while a Nernst lamp illuminated 

 the mirror. 



Witli this form of apparatus the freezing-point of 2'5 c.c. to 5 c.c. of a 

 liquid could readily be determined with considerable accuracy to 0'01° C. ; 

 and a number of determinations of the freezing-points of saps extracted 

 from various leaves were obtained. 



It will be noted that in this arrangement continuous leads connected the 

 junctions with the terminals of the galvanometer ; and at the outset it was 

 thought that, by thus avoiding all discontinuities in tlie circuit, and by having 

 the continuous leads themselves forming the junctions in the couple, all 

 extraneous thermo-electric effects, except those at the terminals of the 

 galvanometer, would be eliminated. This hope seemed to be confirmed 

 wlien, on placing the two junctions close together in a vessel of water, which 

 was vigorously stirred, the spot of light from the galvanometer-mirror came 

 to rest at the scale zero. Greatly to our surprise, however, when the two 

 junctions were put in melting ice, the zero was not attained if any part of 

 the V-shaped couple remained outside the ice and at a higher temperature. 

 This failure to reach the zero was not apparently due to heat conducted 

 more freely to one junction than to the other, as it occurred even when both 

 were immersed precisely to the same depth in the ice, and consequently 

 would receive sensibly the same quantity of heat conducted in from the 

 outside. As soon as the whole of the V was immersed in a vessel at 0° 0., 

 this zero-error disappeared. It seems that for such a sensitive galvanometer 

 the bend acts as a junction, and the two halves of the nickel, not 

 being absolutely identical, act as elements of a couple. The difference of 

 temperature of the air and the melting ice is able, with tliis unlooked-for 

 couple, to produce a deflection of tlie spot of light on the scale amounting 

 to 4-7 mm. 



