CANE SUGAR. 



123 



determined. We were strongly inclined to ascribe it in the main, if 

 not altogether, to inversion. But why should so much inversion occur in 

 Series II when so little of it had been detected in Series I by the method 

 of Fehling? Admitting the greater accuracy of the optical method, it was 

 not possible that so much invert sugar should have escaped detection in 

 Series I, especially since the fault of Fehling's method is its liability to 

 overestimate rather than underestimate the products of inversion. 



Table 15. — Cane sugar, Series I and II. Extreme variations in bath temperature. 



The explanation which sufficed for a time was plausible. It was 

 at the beginning of the work in Series II that the penicillium pest 

 made its appearance, and it was reasonable, as we then thought, to 

 ascribe the apparently greater inversion in Series II to an infection 

 of the membranes by penicillium. The mistake in practice which was 

 made was, of course, in wholly discontinuing for a time the use of 

 Fehling's test as soon as the polariscope became available. An exami- 

 nation of the solutions of Series II by both methods would have shown 

 at once the inconsistency of our interpretation of the loss in rotation. 

 It was probably true, however, that the penicillium did cause some 

 inversion in the solutions of Series II, though by no means enough to 

 account for the whole, or any larger part, of the loss in rotation; for 

 it was found in later series — where the Fehling test was again applied, 

 and after the solutions and cells had been habitually treated with 

 all the care necessary for the suppression of penicillium — that the 

 evidences of inversion had nearly disappeared. 



A circumstance which at the time tended to strengthen the impres- 

 sion that the loss in rotation was due to inversion was the molecular 

 weight which was derived from the osmotic pressures after correcting 

 them for inversion proportional to the observed losses. The mean 

 molecular weight thus obtained was 337.59 (H = 1) instead of 339.60, the 

 theoretical value. The mean molecular weight derived from Series I 

 was 341.41 (0 = 16) instead of 342.22. 



