INTERNAL TEMPERATURE OF LEAVES. 233 



heated by the sun's radiation it gave out reflected thermal 

 rays, which may have affected the leaf temperature. In the 

 new form of the apparatus the leaf was placed with its petiole 

 in water in a small beaker, while the upper end of the lamina 

 was fixed on the hooked ends of two glass rods projecting from 

 a retort stand. By this means the leaf was suspended so as to 

 be freely exposed to the air on all sides, and so asjbo be raised 

 some eighteen inches above the wooden stand on which the 

 whole apparatus was placed. The whole apparatus was 

 shaded behind a wooden screen, in which a square hole was 

 cut, which could be closed to shade the leaf or opened to illu- 

 minate it, by a movable shutter. In this way no part of the 

 apparatus but the leaf was illuminated, and errors due to 

 reflected radiation from the parts of the apparatus were 

 reduced to a minimum. 



The mercury cups were cut in corks held in position by 

 clamps attached to the retort stand, to which the funnel and 

 glass rods and thermometer were also attached. In all cases 

 the leaf was placed approximately normal to the sun's rays. 

 Absolute accuracy in this particular was not always attained, 

 especially in the case of thin flaccid leaves. The position, 

 however, was never departed from to any considerable 

 extent. 



The thermo- junction, whose temperature was under control, 

 was placed inside a rubber tube and close to the bulb of a 

 thermometer w^hich was immersed in water in the funnel 

 attached to the retort stand. A glass rod was used as a stirrer 

 to ensure an even temperature throughout the mass of water 

 in the funnel. 



The galvanometer used was a direct reading suspended coil 

 ammeter, manufactured by R. W. Paul, London. Its resist- 

 ance was 9*2 ohms, at 16° C. This instrument was found 

 very convenient, as it could be carried about with the apparatus 

 to different parts of the gardens and used on the spot where 

 the observations were taken. Each scale division corresponded 

 with the apparatus set up, as usual, to 5° C, and since ig of 

 a division could be easily read off, it was possible to read with 

 accuracy to • 5° C— an accuracy more than sufflcient for the 

 purpose. 



