RADIATION AND TEMPERATURE MEASUREMENTS. 33 



iron-constantan were employed. Two types of couples were tried. The 

 couple shown in Fig. 13, A, consisting of wires 0.005 mm - m diameter (the 

 constantan wire being about 3 cm. long) attached to heavy copper wires 

 mounted upon a wooden base, proved the more satisfactory. In this instru- 

 ment the junctions were fine globules of solder hammered flat and black- 

 ened. In the second type the iron wire (0.2 mm. in diameter) was filed to 

 a point to pierce the outer integument. The great thickness of the iron 

 introduced heat conductivity, which sometimes introduced irregularities, 

 either heating or cooling, when only one of these effects was supposed to be 

 present. It was difficult to operate and hence was discarded for the couple 

 made of fine wires just mentioned. The latter was laid on several folds of 

 cheese-cloth, which reduced the effect of air-currents. The experiments 

 were carried on in a well-darkened room. The effects observed being so 

 marked, it was necessary to reduce the sensitivity of the galvanometer (half 

 period usually 1 second; /'=7Xio~ to ampere) and keep a resistance vary- 

 ing from 200 to 700 ohms in series with it. 



The insect was attached to a thin wooden rod (15 cm. long, and held hori- 

 zontally) in such a manner that it could be made to touch either one or both 

 of the thermo-junctions. Covering either one or both of the junctions by 

 means of a bit of cork or a dead insect, attached to the rod in the same 

 manner, produced no effect. It is therefore evident that the thermal meas- 

 urements made on live active insects are realities and not due to instrumen- 

 tal errors. Furthermore, the galvanometer deflections were decisive, not a 

 few millimeters, but usually from 15 to 20 cm. and frequently " off the scale." 

 The results are not based upon a series of measurements in which the mean 

 of the positive deflections outweigh the mean of the negative deflections, or 

 vice versa. 



While the magnitude of the deflections varied enormously, owing to poor 

 contact, one could continue for 10 to 30 minutes at a time with rarely a false 

 reading. For example, starting with the first or second abdominal (the 

 luminous) segment, Fig. 1, touching the upper thermo-junction, Fig. 13, A, 

 the galvanometer would indicate a cooling of this junction which was usually 

 "off the scale." Rotating the insect about a vertical axis through 90 , with 

 the luminous segment over the junction and with the third (adjacent) dark 

 segment on the lower junction, the galvanometer would go off the scale in 

 the opposite direction, indicating that the upper junction is now the hotter. 

 In other words, while the insect is cooler than the air in the room, the dif- 

 ferential measurements show that the luminous segments are hotter than 

 the dark segments. Reversing the operation by starting with one of the 

 luminous segments touching the lower junction (cooling), then with one of 

 the dark segments on the upper junction, the galvanometer deflection indi- 

 cated that the lower junction was now the hotter. 



Sometimes only the heating effects would be tested, starting, say, with a 

 luminous segment on the upper junction and a dark segment on the lower 

 junction, then the reverse, and the test would be kept up in rapid succession 

 for 10 to 15 minutes, when another insect was tried. In this series there 

 should always have been a heating. Sometimes there would be a cooling 

 effect and, on examination, it would be found that the dark segments were 

 not touching the thermo-junction. While such a series was in progress, and 



