404 



For the I'eaiiiiig at X 6443 no probahle eiToi' is given as only one 

 reading was made for that coionr. The agreement of the experiments 

 with the formula of Lokkntz is evident. 



In Fig. 5 I have represented graphically the residts obtained. For 

 A 4500 and ;. 4580 the theoretically and experimentally determined 

 points coincide. Perhaps it is interesting to give also the values of 

 Fresnkl's coefticient f : 



4500 0.443 0.4(}4 0.465 



4580 0.44'2 0.463 0.463 



5461 0.439 0.454 0.451 



6870 0.435 0.447 0.445 



Here f/,, = I — ^^, f /^ = 1 — _ ^ '1' and f,,,^, ,s found from 

 the numbers in the fourth column of the table concerning the shift of 

 the interference fringes (under A„^,) by multiplication by ^"^ . 



A few words uiay be said conceriung the determination of the mean 

 velocity 'r„ = 465 cm/sec, /j = 214 k.g./cm^ which was important 

 for the interpretation of our observations. We have mentioned already 

 that there was a watermeter in the main tube. This meter (of the 

 WoLTMANN-type) ran very regularly, so that no vibrations were 

 transferred to the system of tubes. It was destined however for 

 large quantities. Its errors were knowji in rough approximation only. 

 If the meter was supposed to indicate accnrately, we found taking 

 into consideration the above mentioned precaution (see p. 401) concern- 

 ing the reading at a complete rotation of the nands of the counting- 

 piece. //'„ = 475 cm/sec, p = 2:\4. k.g./sec. With this value I found 

 a ditïerence of about 2.1 "/„ between the results of ray experiments 

 and the fornuda of Lorentz. In order to investigate, whether this 

 difiference might be ascribed to an error in the watermeter, I decided 

 to put a more accurate measuring apparatus at the end of the system 

 of tubes to control the first watermeter. With extreme kindness 

 Mr. Ing. Prnnink, Director of the Amsterdam waterworks put at 

 my disposal a calibrated so-called "Ster" meter, which begins to 

 indicate at a quantity of 10 L. per hour and which indicates accu- 

 rately for 30 L. and more per hour. If this "Ster" meter was con- 

 nected to the end of the system of tubes, while the principal cock 

 was (pnte open, the mechanical vibrations of the systems would 



