( 36 ) 



it appears that series a gives the lowest results for T m , series b the 

 highest. The fact that the values for a are lower than for c, must 

 probably be ascribed to the circumstance that with almost equal in- 

 ternal bores the wall is thickest for the first series and hence the 

 chance of abnormal stresses is greater. For .series b the wall is 

 thicker than for c, but the inner bore is much smaller, and hence the 

 existence of inequalities and scratches on the surface which unfavourably 

 influence the breaking stress x ) are reduced to a minimum. For the 

 tubes for which '2R' = 1 mm. it seems that the two factors neutralize 

 each other, for those with the smaller inner bore the favourable 

 influence of the surface being smaller preponderates. 



In order to investigate in detail in how far the above mentioned 

 two unfavourable factors influence T m we have applied the direct 

 determination with thin glass threads of which the surface is as 

 smooth as possible and where owing to the small bore abnormal ten- 

 sions are necessarily small. The results which are much higher than 

 those of Winkelmann and Schott, agree with those found by means 

 of the first method and seem to justify the supposition made above 

 about the unfavourable influence of a not perfectly smooth surface 

 and inner abnormal stresses. They point to an upper limit for 

 P m = 1700 atms. 



§ 5. Determinations at low temperatures. 



The determinations in liquid air were made in the same way 

 as those at ordinary temperature. The lower hook of the glass 

 thread was fastened to a wooden bearer, placed beside the thread 

 in a vacuum glass with liquid air. The first determinations gave 

 results which differed much from the later ones. Their mutual 

 agreement is very bad and they are characterized by very high 

 values for the maximum strain, which vary from 44 to 73 KG. 

 per mm. 2 while for the ordinary temperature the highest strain was 

 23 KG. per mm. 2 . Also the structure of the section was totally different, 

 being scarcely ridged but smooth. The smooth spot on the section was 

 as a rule missing. In these measurements the threads were pulled 

 asunder almost immediately after they had been placed in liquid air. 

 Before the following measurements they were left at least 20 mi- 

 nutes in the bath of low temperatures. The latter gave lower results 

 with a better mutual agreement. The structure of the section is 

 similar to that at ordinary temperature, generally a little less distinct. 

 The results are combined in the following table. 



l ) Gf. Winkelmann and Schott, loc. cit. 



