ON A QUAKTZ THREAD GRAVITY BALANCE. 227 



clear that the spring must satisfy sorrie peculiar conditions. The thread must be 

 stretched by a constant, or nearly constant force, and the spring, while freely 

 allowing this to occur, must be incapable of moving so that the point of attachment 

 of the thread shall become displaced transversely. The spring must also be subject 

 to so much damping that it will not vibrate during transport, otherwise the rate of 

 subsidence of the thread may be affected and errors introduced. We have attempted 

 to meet these conditions in the following manner. The spring, as shown in the 

 drawings (Plate 13), consists of a pair of elliptical springs crossing each other at right 

 angles, the point of support for the thread being at the centre of the crossing. The 

 point to which the thread is attached is very light and might be lighter with advantage, 

 so as to increase the effect of the damping. 



There are four bars of spring steel (watch main-spring, in fact), which are attached 

 to the plate carrying the coach springs so as to stand up perpendicular to the plate. 

 The free ends of these springs are attached by wire links to the thread support. The 

 object of this disposition is to damp down the free transverse vibrations of the 

 support. For a long time this formed the complete apparatus, which will be referred 

 to hereafter as the " rosette " spring. 



In October, 1898, however, we had reason to believe that some small residual 

 irregularities were traceable to a transverse movement of the thread attachment. 

 We therefore put on three stays, made of very fine glass hairs, so as to hold the support 

 to the three girders of the main framework. The stays were cemented in position by 

 paraffin, and the wire linkages were also cemented at their points of contact by 

 means of paraffin. This has been completely successful, as the observations will show. 

 We are, however, of opinion that we could improve this part of the apparatus still 

 further, for paraffin is not an ideal cement. In fastening the thread up to the 

 support, and afterwards in stretching the thread, it is very convenient to have a 

 slow adjustment. This is supplied by fastening the spring system on the end of a 

 bar ; the bar is passed through the end bearing and is adjusted by means of a nut 

 working on a thread cut on the bar. The proper tension of the thread having l>een 

 arrived at, the nut and bar are soldered up to the end bearing. In view of the great 

 difficulty of preparing a good thread, it is well worth while to provide every possible 

 convenience for adjustment, and for minimising the risk of breakage. 



We believe that we wasted some time in endeavouring to obtain consistent results 

 with mercury-in-glass thermometers. We were very unwilling to adopt platinum 

 thermometry on account of the loss of portability which is its inevitable concomitant. 

 Through the kindness of Mr. GRIFFITHS we obtained three of TONNELOT'S ther- 

 mometers, which were studied very carefully by M. GUILLAUME before they were sent 

 to Mr. GRIFFITHS. They were again studied by Mr. GRIFFITHS before they were sent 



2 o2 



