188 



SCIENCE 



[N. S. Vol. XLIII. No. 1102 



making hypotheses, or in formulating gen- 

 eralizations, could not have produced re- 

 sults of like significance. It is to be hoped 

 that the probable limit attainable by the 

 present method of construction may be 

 reached, and that we may soon have a 20- 

 inch grating with a resolving power of 

 more than a million. 



A technique for making metallic replicas 

 of gratings is highly desirable and should 

 be attempted. The few good gratings that 

 can be obtained only after years of pains- 

 taking preparation should be indefinitely 

 reproduced. 



Turning for a moment to the subject of 

 temperature, we find that the first ther- 

 mometer was devised by Galileo and con- 

 sisted of a glass bulb with an attached tube 

 whose end dipped into water. It meas- 

 ured temperature changes with an accu- 

 racy far greater than our heat sense could 

 estimate it, and was truly a wonderful in- 

 strument. It made the sense of sight serve 

 the function of the heat sense, and did it 

 better. The original instrument was af- 

 fected by changes of atmospheric pressure 

 and had an arbitrary scale. These defects 

 were gradually overcome. The bulb was 

 filled with water and the tube sealed. The 

 present fixed points, after many others had 

 been tried, were finally adopted. Mercury 

 was selected as the most suitable liquid for 

 general purposes. The material of the bulb 

 has received due attention, and many modi- 

 fications of the thermometer for various 

 purposes have been devised. It is doubtless 

 the most common scientific instrument in 

 use. The development of the mercury 

 thermometer has made itself felt in every 

 line of research. 



Other means for measuring temperature 

 have been devised. Resistance thermom- 

 eters, thermocouples, bolometers, and a vari- 

 ety of radiation pyrometers, have made 



possible investigations beyond the reach of 

 the mercury thermometer. 



The development of the nitrogen ther- 

 mometer and of the thermodynamic scale has 

 placed temperature measurements on a still 

 more scientific basis. The thermodynamic 

 scale has quite recently been extended to 

 1,550° Centigrade; and all measurements 

 beyond that are still extrapolations based 

 on the law of the thermocouple up to the 

 melting point of platinum (1,775° C), and 

 on the two laws of radiation for higher 

 temperatures. 



The range of temperatures at our con- 

 trol for the study of natural phenomena 

 extends from about two degrees Centigrade 

 above absolute zero to 4,000 degrees, the 

 outer limits being attainable through the 

 invention of the liquid air machine and 

 the electric arc. The oxidation of all mate- 

 rials at high temperatures has made the 

 use of the electric arc impossible for most 

 purposes. A recently devised furnace has 

 overcome this difficulty, enabling experi- 

 ments to be made in any gaseous atmos- 

 phere up to 1,600 degrees Centigrade ; and 

 also produces any temperature nearly up 

 to that of the electric are. The bombard- 

 ment by cathode rays gives promise of the 

 development of extremely high tempera- 

 tures for experimental work in a vacuum. 

 The attainment of the lower limit of tem- 

 peratures has made possible the most won- 

 derful discovery of "superconductivity," 

 which, with the investigations on conduc- 

 tivity at high temperatures, adds most sig- 

 nificant data toward the development of 

 the theory of electric conduction. 



The bellows, the siphon, the water pump, 

 the fact that water is supported in a filled 

 inverted bottle when its mouth is in water, 

 and various other phenomena, were ex- 

 plained on the principle that "nature ab- 

 hors a vacuum." The invention of the 

 barometer by Torricelli (1643) almost 



