100 PROCEEDINGS OF THE AMERICAN ACADEMY 



The shell-like space between steel tube qq and glass tube tt is filled 

 with a liquid of small coefficient of expansion. I first tried mercury, 

 but found it almost impossible to fill tt in such a way as entirely to 

 exclude air. It is clear, in view of the friction of the mercury thread 

 m ss, that any trace of air in tt will vitiate the experiment. The 

 thermal expansion of colored alcohol is too great. I therefore used 

 water containing a little alcoholic solution of fuchsine. This is in- 

 troduced through the capillary, by evacuating the tube tt. It is ne- 

 cessary to repeat the operation a number of times, and to boil out 

 traces of air. I obtained the requisite constancy of temperature, by 

 surrounding tt with a jacket, uu, of circulating cold water coming 

 directly from the hydrant. To put the liquid in ss at a given fiducial 

 mark, a special adjustment for moving the meniscus suggests itself; 

 but this is an exceedingly difficult device to apply, seeing that there 

 must be a minimum of water in tt, and that absolutely tight joints 

 are essential. Hence I raise or lower the merriscus, and color it when 

 faded, by inserting filamentary glass suction tubes into the canal ss. 

 There is then little difficulty either in adding more liquid to the thread, 

 or in withdrawing liquid from it. Cf. § 14. The steel tube qq should 

 be protected against rust, by nickel-plating it. I found, moreover, 

 that water free from air and containing a little alcohol scarcely attacks 

 steel ; for bright surfaces remain bright after many months. 



By using a long steel tube the effective length can be more accu- 

 rately stated, for the length error at the joint tq is jiroportionately 

 decreased. Again, since the expansion corresponding to a given 

 pressure increases in absolute magnitude with the length of qq, a 

 wider capillary, ss, suffices for measurement when qq is long. This 

 facilitates the adjustment for fiducial zero already given. 



14. Tait Gauge. Graduation. — To accomplish this I compared 

 the gauge with a large Bourdon gauge* reading from to 1,000 

 atmospheres. § 1. The tube of the latter communicated with the 

 barrel, through one of the screws M. This comparison is a check on 

 both instruments, and the statistics are therefore given below in de- 

 tail. It affords no means of testing the value of the standard atmos- 

 phere employed; but since both gauges are based on Hooke's law, and 

 provided with suitable scales of equal parts, the relations are well in- 

 dicated. In the table, '2 A^, 2 A^, and L denote the internal and the 

 external diameters and the length, respectively, of the steel gauge tube 

 ' qq. Figure 4 ; 2 p is the bore of the capillary tube ss. 



* Furnished by the Societe Genevoise. 



