Dixon and Atkins — On Osmoiic Pressure in Plants, cVe. 281 



set lip. It is a very slow change, and may amount to 2-3 mm. in a day. 

 The second is only apparent when the circuit is closed, and seems to be due 

 to the difference in temperature of the terminals of the galvanometer. 

 When tlie galvanometer is carefully screened from rapid fluctuations in 

 temperature, this second error is also insignificant during the time of an 

 observation. In cases of excessive fluctuation of temperature it may amount 

 to 10 mm. in the day. In the later observations it was, however, completely 

 eliminated by the use of a reversing key in the circuit, by means of which 

 the deflection due to a difference in temperature of the junctions might be 

 instantaneously changed from side to side of the zero. The mean of the 

 deflections will then give the true deflection, irrespective of a zero-error due 

 to the difference of temperature of the terminals. It is evident that the 

 reversing key used with this object must be of special construction, so that 

 it shall not itself introduce thermo-electric effects into the circuit. "With this 

 object a modification of a key, used already successfully with thermo-couples 

 in tlie investigation of the temperature of plant organs, was adopted. This 

 key consists of a wooden-spring clip, carrying the exposed ends of the leads 

 coming from the galvanometer on the inside of its jaws. The leads from the 

 couple terminate on each side of a glass support. When the clip is closed 

 upon this support, the leads coming from the galvanometer lie at right angles 

 across those coming from the couple. Reversing the clip from side to side 

 of the support reverses the current. Even when the leads coming from the 

 galvanometer and from the couple are made from the same piece of wire, 

 differences of temperature on the opposite sides of the support and the clip 

 will be found to give rise to thermo-electric effects in the circuit. To 

 equalize the temperature of these parts, it was found convenient to immerse 

 the clip and the support in a vessel of petroleum which was kept stirred 

 during the observation. With this device the zero-error was satisfactorily 

 eliminated, and no thermo-electric error introduced. 



The calibration of the apparatus was carried out by one of two methods. 

 At first the incompletely jacketed couple was used. While one junction was 

 kept in melting ice, the other junction, along with a Beckmann thermometer, 

 was placed in succession in solutions of sodium chloride of increasing 

 concentration. The deflection registered by the spot of light, and the 

 simultaneous reading of the thermometer at the freezing-point of the 

 solution, were recorded. By a series of these observations a satisfactory 

 calibration was effected. 



The following are the observations for the calibration of the thermo- 

 couple No. 1 by this method. 



Witli tlie Beckmann thermometer used, the scale-reading 4*140 



