164 BOTANICAL GAZETTE [September 



tube 1°^ l<^i^g f*^^^ of water, whose lower end was in a mercury 

 bath. This was done at 3.40 p. m., January 19. The height of 

 the mercury was : 



At the bottom 4«i from the top 



ram A r r^mm 



January 19, 7:20 P. M., 9.0'"'" 450 



January 20, 9:50 P. M., 28.5 530 



The suction at the bottom was then 5 7°'°'. At i p. m. the 

 connection at 4"^ had sprung a leak. It was unfair to compare 

 tubes of such different caliber, so the U tube w^as replaced by a 

 capillary of the same bore as the other, as nearly as possible, 

 and arranged like it. Starting at 1:15 p, m., January 20, the 

 height of the mercury was : 



At the bottom 4^1 from the top 



January 20, 3:45 p.m., 130"^^^ 455""™ 



6:25 P. M., 186 494 



g:io p. M,, 252 531 



January 21, 10:30 A. m., 326 550 



On January 23 at 10:30 a. m., bubbles having twice been 

 lemoved from the connection, the mercury at the bottom 

 reached 428 



mm 



The experiments continued about two months, during which 

 I took several pages of measurements. At one time I arranged 

 the osmometer so as to record the volume of water it absorbed, 

 and found again that for a limited time it could be withdrawn 

 from the tube very much faster than it was taken up lower down. 

 I tried to determine the difference in the suction at the three 

 openings in the tube when the mercury should be drawn up at 

 each until they were in equilibrium ; but this was reached so 

 slowly that if there was no other accident, air always came out 

 of the main tube and interfered. During the whole time that 

 the experiment lasted there was a gradual increase in the air in 

 the tube, and with this a decrease in the difference in tension at 

 different heights. 



The experiment ended with closing the T tube connections 



and setting the root end in a solution of eosin. As the eosin 



. rose slowly in the tube, it was very easy to see that it left air 



between the plaster and the glass ; rather more air than was 



t, 



■ 



