402 Proceedings of the Hoi/al Irish Academy. 



thus one point on the curve is obtained, the normal temperature of 

 the room (about 12°) being subtracted from the known melting point 

 of the substance used to calibrate. 



A similar process is gone through with each of the other substances 

 used in calibration. I have always found potassium bromide, black 

 copper oxide, and paladium, most convenient. Sometimes a specimen 

 of actinolite was used as coming conveniently between CuO and pala- 

 dium, but with very slight advantage, as the curve was found always 

 to pass very close to the point thus obtained. Actinolite presents 

 considerable viscosity, and thus has no very definite melting point, 

 but still was useful as a verification of the cui've. A numerical 

 example may be useful as to the method of finding the curve of expan- 

 sion for a ribbon. Thus, in the case of black oxide of copper the 

 reading of the screw-head at starting was 1986 divisions from zero, 

 when the CuO melted ; the head was reading 2329 divisions ; so that 

 2329 minus 1986 gave the number of divisions, the head moved 

 through between the normal temperature of the room (12°) and the 

 melting point of CuO, so the screw-head advanced 343 divisions. 

 Each of these divisions represents the ^oiuu part of an inch, therefore 

 the screw advanced through if^w parts of an inch ; the ribbon when 

 cold measured 3-83 inches. 



Dividing 343 by 38,300, the number 0-008955 is obtained. This 

 is plotted on the ordinate to a convenient scale of 0*002 to an inch. 

 Logaritlimetic paper can be procured ruled to this scale. The known 

 melting point of CuO, 1055°, minus the temperature of the room, 12°, 

 was plotted on the ordinate at right angles to the temperature scale 

 at the point corresponding to 1043° C. CuO is then marked on the 

 curve at the point corresponding to 1055° C, the temperature at which 

 CuO is known to melt. The first melting in the case of tliis sub- 

 stance, CuO, must always be used as any subsequent melting is 

 higher. Other substances used for calibration are dealt with in the 

 same manner. 



A curve once obtained for a ribbon (see Plate V.), the determination 

 of a melting point is calculated from an observed expansion by 

 calculating the value (of I2-I1) divided by li, corresponding to the 

 expansion ; then plotting the ordinate and finding the temperature 

 corresponding to this ordinate, and adding the number of degrees 

 corresponding to initial temperature. 



A great portion of one's time would be taken up plotting curves if 

 the above operation was necessary for each new ribbon that was used ; 

 as a ribbon very soon gets dirty, the melted particles of the minerals 



