PHYSICAL LABORATORY: COMPARISONS OF STANDARDS 479 



standard. This comparison is very laborious, and is not recommended 

 except in very exceptional cases, as more than one comparison should 

 be made to insure good results. 



2nd. Comparison with primary standards which have been compared 

 many times with the air thermometer. This is recommended where an 

 error of y^ is of some importance. 



3rd. Comparison with secondary standards which have been com- 

 pared many times with the primary standards, and not very often 

 directly with the air thermometer. This is recommended in all ordi- 

 nary cases, where an error of yf^ can be tolerated. 



When several comparisons are made, the following intervals will be 

 allowed between the experiments, so that the zero reading may be 

 allowed to return to its primitive value. 



Thermometers heated to 40 C. about 1 week. 

 Thermometers heated to 100 C. about 6 weeks. 

 Thermometers heated to 250 C. about 4 months. 



The latter interval is too small for an accurate return. 



For the exact details of the method of comparison, I must refer to the 

 above mentioned paper on the Mechanical Equivalent of Heat. 



It is advisable in all cases where great accuracy is desired, that a 

 numbers of comparisons be made, seeing that delicate thermometers are 

 constantly varying through slight limits, and the average state can only 

 be determined by repeated experiments. 



Reports. In the report of the comparison, the original readings will 

 be given together with the reduced ones, and the plot of the curve of 

 errors of the thermometer at every point. From this curve, the error 

 of the thermometer at any reading can be found. 



It is proposed to publish at the end of the year a complete report of 

 all the comparisons made during the year, together with all new deter- 

 minations of the errors of the standards, and to send it to any address 

 at a price which we will hereafter announce. 



Fees. The comparators allow five thermometers only to be placed in 

 them, of which two are our own standards in ordinary comparisons, 

 and one in direct comparisons with the air thermometer. Therefore, 

 three thermometers can be compared as easily as one in ordinary cases, 

 and four in direct comparisons. Hence the following system of fees 

 has been made out. 



