790 REPORT— 1898. 



Righi has also observed the converse effect in nitric oxide. A small tube of 

 £^lass, 32 mm. long and 15 mm. in diameter, is closed at its ends with thin discs of 

 microscope glass cemented on -with gutta-percha, and is provided with side-tubes 

 by means of which it can be filled with dry nitrous fumes. This is placed between 

 the poles of the magnet in a field of intensity about 2,000 C.G.S. units. A strong 

 beam of white light is passed through the polariser at one end, then through the 

 tube containing the nitrous fumes (in a direction parallel to the magnetic lines), 

 then through the analyser, the latter being turned so as to cut off all the light. 

 Looking back through the analyser the field appears absolutely dark. On turning 

 on the exciting current there instantly appears a green-blue light, this being the 

 tint which is complementary to the yellow-red that nitrous fumes usually show by 

 transmission. 



Applying a hand spectroscope one observes that the spectrum of this gpreen- 

 bliie light is the complementary of that of the ordinary absorption spectrum ; 

 though in detail each dark ' line ' would be replaced by two bright ' lines ' close 

 together. In fact, for every ray of frequency w, that would be absorbed by the 

 substance -when not in the magnetic field, there issue two rays of slightly difi'ering 

 frequencies w., and w, ; so that the light which appears when the field is made is 

 practically identical (unless very great dispersion is applied to resolve it finally) 

 •with the light absorbed, however that light may be distributed in the spectrum. 



If while looking at the green-blue light through the spectroscope the analyser 

 is slightly turned, we get the ordinary spectrum of the yellow-red light due to 

 absorption. 



Here, then, we have the curious result that we can observe the emission-spectrum 

 of a non-luminous gas. 



This has been confirmed in the following experiment. A peUet of metallic 

 sodium was gently heated in a current of dry hydrogen so that a non-luminous 

 vapour of sodium was produced in a tube lying along the magnetic field. The 

 arrangements were otherwise as before. When the analyser was crossed no 

 light wliatever was seen until the magnet was excited, when the ordinary radiation 

 spectrum of sodium vapour (modified in reality by the doubling of each line) at 

 once flashed out. 



10. On the Dissipation of Energy in the Dielectric of a Condenser. 

 By Edward B. Rosa and Arthur W. Smith. 



The heating effect of an alternating E.M.F. upon the dielectric of a condenser 

 is measured by a wattmeter, using a coil of wire in series with the condenser to 

 give resonance and raise the voltage upon the condenser. The wattmeter measures 

 the energy expended upon coil and condenser, the current being nearly in phase 

 with the E.M.F. The coil loss is then deducted from the total to give the net 

 energy expended upon the condenser. This is then divided by the energy stored 

 to give the net loss, in per cent., and hence the efficiency. 



The efficiency is defined as follows : — 



_ Energy stored — Energy dissipated 



V- 



Energy stored 



and T) is proved to be 1 — tt cot (^, where <f> is the angle between the current and 

 E.M.F. for the condenser. 



In a second series of experiments the condensers are placed in a calorimeter, 

 especially designed for the purpose, and the heat measured while they are subjected 

 to a measured E.M.F. at a known frequency. This gives the efficiency, which 

 agrees with the previous work, though by a totally different method. 



It is found that the percentage of loss in paraffin paper condensers is relatively 

 small, varying with the temperature and condition of the paper before paraffining 

 it. In fjee.nvax and rosin condensers the energy lost is much greater, and increases 

 as the temperature rises up to a maximum value and then decreases as the tempera- 

 ture rises further, until the dielectric begins to soften, then it increases rapidly. 

 Exact numerical values are given for the losses. 



