( 108 ) 



indicated in § 3 of Comm. N°. 117 (June 1910) with the mercury 

 at the same pressure in the two communicating vessels, the manometer 

 and the volumenometer. For this purpose the upper portions of the 

 volumenometer and the manometer were brought into communication 

 with each other and the mercury was then driven up until the 

 meniscus stood in one of the connecting necks of the volumenometer; 

 the clamp 4 (PL I fig. 1, cf. Pi. T, Comm. N". 117) was then closed 

 so as to avoid the effects of vibrations in the mercury in the bulb 

 (2i (cf. PI. I Comm. N". 117) caused by gusts of wind or by vibration 

 of the building. Care was always taken to ensure a good vacuum 

 above the two corresponding mercury levels, so as to prevent trouble- 

 some expansions and too slow equilibration between the gas in the 

 volumenometer and the manometer. Since in the case of the uppermost 

 necks mercury columns about 1 metre in height had to be kept in 

 hydrostatical equilibrium, the greatest care had to be taken to ensure 

 the temperature equilibrium of the mercury colimin. For this purpose 

 a water jacket was put around the manometer tube being firmly 

 attached to it by means of two rubber stoppers. In order to reduce to 

 a minimum the optical corrections necessarily introduced by the use 

 of this jacket, the holes in the rubber stoppers were bored to one 

 side of the centre so that the water jacket was thinnest just at the 

 side through which the readings were made. In order to determine 

 the optical corrections fifteen fine lines were etched on the other 

 side of the manometer tube. The heights of these lines were also 

 read on the cathetometer with the volumenometer jacket both empty 

 and full of water. Each of these measurements was repeated sixteen 

 times. Since both the water layer and the air layer through which 

 readings were made w^ere only 4 mm. thick the cathetometer telescope 

 could be kept immovable and the difference could be determined 

 directly each time with the micrometer eyepiece. Parallax errors 

 were avoided by cosering one half of the telescope objective with 

 a plate of glass whose optical thickness was the same as that of the 

 water layer in the volumenometer jacket. When the jacket was full, 

 readings were taken through the uncovered half of the objective, 

 and when it was empty readings were made through the covered 

 half. The optical error due to the glass plate was determined inde- 

 pendently. In this way the optical errors due to the manometer 

 jacket were easily determined with accuracy to within 0.005 mm. 

 From the results it is evident that the use of such a jacket has 

 much to recommend it, while the optical correction appears to be 

 only about 0,03 mm. a value that may in the majority of cases be 

 regarded as negligible. 



