308 PROCEEDINGS OF UNITED STATES NATIONAL MUSEUM. 



itself like a crook, graduated ia fifths and marking from 55° to 74° F. 

 (self -registering). 



5. A Negretti-Zambra deep-sea thermometer, graduated in degrees 

 only, and ranging from about 25^ to 100° F. 



All of these excepting the Negretti-Zambra are graduated upon the 

 stems. The three highest in range of lot 2, one of the long standards, 

 and the crooked instrument were made at first self-registering, on the 

 principle of clinical thermometers, by a break in the column of mer- 

 cury. They were so ordered in the hope that it would be possible to 

 make some of the experiments upon living fish in tanks where the water 

 could be artificially warmed above the temperature of the air. Such 

 experiments not being possible (for reasons known to you) the self- 

 registration was destroyed by reuniting the broken mercury column, 

 and the necessary small correction applied. The curved thermometer 

 was intended for use in a living fish, the bulb to be inserted either into 

 the rectum or into an incision in the muscular tissue, and the stem to be 

 secured to the body of the fish, which was then to swim free in the water. 

 The highest temperature reached would be registered by the thermom- 

 eter. This instrument, like the other self-registering thermometers oper- 

 ating on the same principle, can only be made available when the tem- 

 perature of the water is above that of the air, and there has, therefore, 

 as yet been no opportunity to make use of it. 



Owing to the curious molecular change which occurs in the glass of 

 which thermometers are made, whereby, after from six months to a year, 

 the instruments show an error of excess of from half a degree to a degree, 

 these thermometers, which were necessarily "pointed" as soon as made, 

 are not strictly accurate. They should be returned to the maker and 

 rated again before being used next summer, so that the necessary cor- 

 rection may be applied. For the time being the error has been to some 

 extent met by Mr. Tagliabue, who has " overpointed " the scale about 

 half a degree. I would also suggest the propriety, in case you conclude 

 to continue temperature-observations, of ordering in advance one or two 

 long thermometers, marking from 30° to 100° F., to be " pointed " after 

 six months and used as absolute standards. Since, however, in these 

 observations, absolute temperature is less important than relative accu- 

 racy, I have taken much pains to rate all the instruments together, 

 comparing them with the standard, and applying such corrections as 

 will reduce all the readings to its scale. The same error, if any, will 

 then be present in all observations, and relative accuracy will be pre- 

 served. In Table A, which contains the corrections deduced from more 

 than three hundred separate comparisons taken at nearly every degree 

 on the scale, the small thermometers in daily use are numbered from 1 

 to 5 for the first set, and from 6 to 12 for the second set (which has not 

 yet been used). No. 1 being the thermometer of lowest scale. The com- 

 parisons were made by immersing the thermometers in water, artificially 

 cooled or heated. Only the means of each 10° are given in the table. 



