THE Sl'HdCTirKE OK TIIK NIICMOUS. ]55 



A few special observations were made witliout partitions. Coiivective vortices 

 due to the waiiiier glass walls after exiiaustioii should Ije upward ou the outside 

 and down in the axis. The initial strands of fog from nuclei in water vapor are apt 

 to run upward iu the axis and downward ou the outside, according as the nucleation 

 is added below or above, I'espectively. Tiiis peculiar and contradictory behavior 

 is difficult to explain. The whole appearance is something like that of a burning 

 granule of powder throwing off smoke from an advancing center, and it is not 

 impossible that coarse nuclear matter is being disintegrated in some similar way as 

 the nuclei in water va})or are ultimately very small. 



Experiments ti'ied with formic acid wei'e all I'ailui'es from the invai'iable 

 presence of water vapor. 



INFERENCES AND SUMMARY. 



17. Hates of diffic^ion. — The irregular distribution of results does not suggest 

 a suitable curve to be passed through them. The heights obtained after the lapse 

 of long intervals of time frequently indicate a retarded rate ; but this is to be 

 otherwise expected seeing that after long intervals the numl)er of particles in the 

 plane of diffusion must continually decrease as a result of the losses laterally, and 

 eventually the true plane of ditt'usion will fade beyond recognition. This will 

 be particularly the case when water vapor is present, for under these circumstances 

 the enormously rapid dift'usion of the nuclei associated with water vapor will act 

 as an additional and continued drain on the nuclei associated with the other vapors. 

 It is from this point of view that the al)normally low values in the tables are to be 

 explained. 



The effect of temperature, though pi'esumaljly of marked importance, is beyond 

 recognition in the case of methods of the kind here discussed, and it was there- 

 fore ignored. 



For the reasons indicated, it will suffice the present purposes to draw mean 

 straight lines through the observations, or, in other words, to consider theii- avei'age 

 distribution and trend. 



In table 12 the results for the absorption velocity, h in cm. per minute, are 

 given in full. The mean rise of the plane of demarcation per minute as at once 

 obtained from the graphs, is entered under h'. This is to be corrected for the 

 instantaneous rise due to the exhaustion by multiplying it by the pi-essure ratio, 

 p'/p; h is so obtained. All the graphs point to an initial height (time t — 0) of 

 the plane of demarcation between 5 and 15 cm., depending on the means available 

 to depress it. There is good reason for this observation, as the nuclei can not be 

 so slowly and carefully introduced as to insure complete freedom from motion. If 

 the initial height were ignored it would be necessary to assume a sharp initial 

 curvature for all the graphs, a condition of things much less wari'an table apart 

 fi'om gravitation. Moi'eover, in the case of benzol, where many observations were 

 made for the same pressure difference, the raised position of the plane of demar- 

 cation was directly demonstrated. It was also shown that fi-om the same charge of 

 nuclei, diffusion increases perceptibly in the lapse of time, due to the gradual 

 accumulation of the smaller particles in the front of the advancing column. 



