26 On the Conduction of Heat in Liquids. 



the liquid, which is surrounded by a metal cylinder at T °; the 

 liquid before entering the cylinder having a uniform tempera- 

 ture T 1? and leaving with a mean temperature U. Now if 

 Graetz be correct in supposing that the metal cylinder is so 

 good a conductor as to be of the temperature T throughout 

 its entire thickness when surrounded by water of temperature 

 T , it seems obvious that the ends of the pipe cannot preserve 

 a totally different temperature. For instance, at the end 

 where the liquid enters, the part of the pipe imbedded in the 

 cork seems at least as likely to have the temperature T as the 

 temperature T x ; and there can be little doubt that in reality 

 the temperature is intermediate between the two, and varies 

 from point to point of the pipe. Since I occurs as a factor of 

 k in the formula, any error in / produces an exactly propor- 

 tional error in k in the opposite direction. If the above view 

 be correct, I is uniformly given too small, and so the value 

 found for k is always too large ; but the error in I would differ 

 from one experiment to the next. Thus the values obtained for 

 k would not lead to correct measures of relative conductivities 

 even. 



In the tables of results Graetz does not directly give the 

 volume issuing from the pipe per minute ; but he gives the 

 value of a quantity t, which is defined some pages previously 

 as the time in seconds required to pass 100 cubic centim. of 

 the fluid through the pipe. Thus we should have V, given 



by the equation V = The substitution, however, of 



this result and the other data in the equation does not seem 

 to satisfy it even approximately; so probably t has some mean- 

 ing to which the paper affords no obvious clue. 



The absolute values given in the second paper are much 

 less than in the first, and, as a rule, slightly smaller even than 

 Weber's, though referring to a higher temperature. Graetz 

 seemed disposed to assign this difference to the radiation be- 

 tween Weber's plates ; but numerical results deduced from an 

 application of Stefan's law of radiation to the data given by 

 Weber indicated that not more than 2 per cent, of the effect 

 could have been due to this cause. He thus attributes the 

 difference to Weber's neglecting the loss of heat by external 

 conduction from the liquid surface. 



On the whole I do not think Graetz's method very satisfac- 

 tory, and am doubtful if much weight should be attached to 

 his results. There are a good many points in the theory 

 open to criticism; and the formula obtained is of such a nature 

 that a small error in the value given for the constant p x by the 



