( 700 ) 



therefore undoubtedly right. But it follows by no means from this 

 that also the mean path is smaller in the first case. For it is t-lear, 



/Ï 





•P 



Fig. 1, 



that among the molecules of the group in question which in the 

 period T reach S, there are some which have their last point of 

 collision close to S, e.g. in D. These molecnles with their verv short 

 path do not count for the group which reach B. On Ihe otliei' hand 

 there will be others whose last point of collision lies so far from !>, 

 e.g. in E, that they can reach B, but not *S' in the interval of time 

 T under consideration ; they must therefore be taken into account 

 for B, but not for S. That the mean path remains unchanged a\ hen 

 we take these circumstances into consideration is l)est seen by thinking 

 the space, in which the last points of collision lie, divided by planes 

 parallel to ]> or S. It is clear that all molecules which have their 

 last point of collision in pai'ts marked by the same number, (Fig. 2) 

 hax'e the same |ialh ; also clear that those points of collision are 

 reu-ulai'lv dis(rilmte<l o\ei- the whole xoliinu^ and that the volume 

 of the parts with the same numbers is the same, bO that also the 



mean patli must be the same. So 

 if the reason of correction given by 

 \ AN DKR Waai.s wei'e the only one, 

 then the formula (3) would strictly 

 apply unmodiiied ; this is, howe\er, 

 not the case, as there is anothei" 

 reason for correction, which we have 

 not taken iido account as yet, as this 

 ^'B- 2. intluence was exj)ressi\ely excluded 



by VAN DER Waals and Korteweg in their calculation, by van der 



CÏ 



