94 On Rotatory Polarization in Liquids. 



indicated by two thermometers passing through side-tubes — 

 one at each end of the tube containing the solution. The- 

 temperature of the solution in all the experiments was about 

 18° C. It must be pointed out that large readings mean 

 small rotation. The position of the polarizer is different for 

 each set of readings, so that there is no relation between the 

 numbers in different sets. This of course does not interfere 

 with a comparison of the arc and spark readings. 





Circle-reading 



Circle-reading 



1 



Circle-reading 



Circle-reading 



Set. 



for arc- 



for spart- 



Set. 



for arc- 



for spark- 





light. 



light. 





light, 



light. 



I. 



96° 59' 



96° 56' 



; II. 



97° 33' 



97° 26' 





54 



53 





27 



15 





56 



97 





25 



23 





57 



4 





30 



32 





57 



2 



96 57 



97 1 



96 57 

 48 





33 



36 

 29 

 23 



III. 



100° 32' 



100° 24' 



IY. 



97° 22' 



97° 36' 





30 



42 





25 



9 





34 



21 





25 



32 





33 



23 





27 



26 





35 



34 





27 



23 





30 



21 





25 



20 





34 



42 





27 



53 





33 



47 





26 



15 





31 



46 





25 



30 





33 



47 





25 



27 



29 

 36 













38 



If all the observations are included in the estimate of the 

 means, the values are as follows : — 



Arc. Spark. 



Set I 



. 96° 



56'-G 



96° 



5 7'- 5 



II 



.. 97 



29-6 



97 



26-3 



Ill 



.. 100 



32-5 



100 



34-7 



IY 



.. 97 



25-5 



97 



28-9 



Grand means . 



. 98 



36-05 



98 



36-8 



The total rotation is about 500° ; hence the grand means 

 differ by less than 1 minute in 500°, or less than 1 part in 

 30,000.' It is noteworthy, however, that in three out of the 

 four sets the rotation is less for the spark than for the arc. 

 In the exceptional set (II, spark) occurs one very low 

 reading — too low to be due to ordinary experimental errors. 



