528 bulletin: museum of comparative zoology. 



flask is g inch greater than that of the thermometer-bulb. There is 

 little danger of breaking the bulb therefore when inserting the works 

 into the flask. The tapered portion of the sleeve in which the ther- 

 mometer is mounted steers the works into a central position, and the 

 shoulder of the sleeve, Plate 1, fig. 1 y, Plate 4, fig. 12 y, rests upon the 

 shoulder, Plate 5, fig. 14 M, of the flask, a tight joint being secured 

 by a rubber washer placed on the thermometer-sleeve. At z, Plate 1, 

 fig. 1, is shown a groove cut in the works, which carries a wire-spring 

 of peculiar shape. This spring holds the upper portion of the works 

 in a nearly central position in the flask, and the spider-spring N, 

 Plate 5, fig. 14, attached loosely to the plug, presses down upon the 

 cover of the works. Supported in this way the works cannot be 

 injured by a slight distortion of the flask. 



The thermometer-bulb projects into the long narrow portion of the 

 flask and good thermal contact is insured by nearly filling the space 

 between them with mercury. If the lengths of the various inter- 

 changeable thermometers be properly calculated, they will all work 

 with the same quantity of mercury in the flask. Since the bulb is 

 in good thermal contact with the sea-water and in very poor contact 

 with other parts of the apparatus, through glass only, the temperatures 

 of the works and upper portion of the flask have little effect on the 

 records. Experiment shows that under the most unfavorable condi- 

 tions the thermometer will come to the temperature of the water 

 closer than .1° in about one minute. 



The lower narrow part of the flask is protected by a guard, Plate 5, 

 fig. 14, to which for work at considerable depths there should be 

 attached, a short wire carrying a heavy weight. 



The thermograph has not yet been tested in deep water. It has 

 been tried in the Charles River and in the laboratory, and has been 

 found to work properly. The records may be read with an accuracy 

 of about .05° when using the range —2° to 33°, and are correct within 

 these limits. If the collimator, Plate 2, fig. 6, contained more plates 

 the accuracy of reading would be increased. There is no reason why 

 one should not be able to read .01° in this way. Until the instrument 

 was constructed it was not known how much exposure the film would 

 require. It has since been found that there is plenty of light to war- 

 rant the use of a collimator with plates much closer together. 



If it should prove desirable to obtain more than 200 exposures on a 

 single film, a slight alteration of the clockwork and a narrow stop 

 placed before the film would make it possible to get about 800 records 

 with one winding of the spring. This would mean a record every 

 three minutes for forty hours. 



