532 



NATURE 



[December 22, 1921 



observe the transit of Venus and to study atmospheric 

 refraction in the tropics. As an incident of the latter, 

 while at Pondichery he observed carefully the form 

 of the rising sun, with attendant phenomena, and 

 noted the instants of upper and lower contacts in 

 several cases, with the help of the telescope of his 

 quadrant and a clock. While he does not seem to 

 have made any further use of the durations deducible 

 from his observations, nevertheless with the help of 

 the Naii>tical Almanac of that year and a standard 

 astronomical formula one may compute the duration 

 which would have held on the assumption of constant 

 refraction, and may compare this with the actual 

 observations and with the accompanying descriptions 

 of the sun's form. The subjoined table shows the 

 results. Type A (see Popular Astronomy, " On Types 

 of Sunrise and Sunset," vol. 29, p. 251, 192 1 ; Nature, 

 October 13, p. 211), which is a mirage type, has an 

 average excess of +23 per cent, as compared with 

 computation ; the one type B case has an excess of 

 + 5-3 per cent. These results agree with numerous 

 observations made by the present writer at various 

 stations in Pacific and Atlantic waters. It is as if 

 the sun sets behind a receding horizon and rises beyond 

 an approaching horizon. 



Duration 



The Measurement of lonisation Currents by Three- 

 electrode Valves. 



In view of the advantages offered by the use of 

 valves for measuring ionisation currents, which have 

 been pointed out in a recent paper by M. J. Malassez 

 (Comptes rendus, vol. 172, p. 1093, 192 1), it appears 

 to be of some interest to describe another method of 

 using valves for this purpose, differing in its applica- 

 tion from that adopted by M. Malassez. 



The method described by M. Malassez gives directly 

 the mean ionisation during a certain interval of time 

 by determining the time required to discharge a con- 

 denser from 40 volts to a small potential indicated 

 sharply by the valve. It is intended to measure ionisa- 

 tion currents, of the order of 10-' amperes, produced 

 by X-rays when conditions are such that the leakage 

 through the valve can be neglected. 



In the method to be described the ionisation is 

 measured by the steady deflection of a galvanometer. 

 It permits changes of ionisation within short intervals 

 of time to be followed, and can readily be adapted to 

 a balance method. 



The arrangement is given by the accompanying 

 sketch (Fig. i). 



The plates of the valves Vj, V^ are connected 

 'through equal resis-tances r^, r,. One terminal of 

 each of these resistances is connected to the positive 

 pole of the battery, the other terminals are con- 

 nected through the galvanometer g. The resistances 

 R,, R, regulate the heating currents through the fila- 

 ments. Each grid can be connected to the positive plate 

 of an ionisation chamber through the leads Cj or C^. 



The heating currents are set so as to give equal 

 plate currents through both valves when the grids are 

 insulated. When, owing to the action of the ionisa- 

 tion current, one of the grids receives a charge, the 

 plate current through the corresponding valve is altered 

 and the galvanometer is deflected. An ionisation cur- 

 NO. 2721, VOL. 108] 



rent acting on the other filament causes a galvano- 

 meter deflection in the opposite direction. 



The sensitiveness is, in general, different for ionisa- 

 tion currents acting on each of the grids, but for each 

 one the deflections of the galvanometer are fairly pro- 

 portional to the ionisation currents. If required, in- 

 equalities in the valves can be corrected by interposing 

 an E.M.F. in one of the valve circuits by means of a 

 battery and suitable resistances. 



When two valves are used in this way, fluctuations 

 in the heating currents and in the plate potentials 

 are reduced owing to the fact that the healing cur- 

 rents for the filaments and the plate voltage are sup- 

 plied from the same sources for both valves. 



Pio. I. 



A fuller discussion of the method and more details 

 of the experiments carried out will be given elsewhere. 



When some radio-active deposit emitting a-rays is 

 gradually brought near to an ionisation chamber of 

 suitable shape it is easy to show the amount of ionisa- 

 tion for different distances (Bragg 's curve) by the 

 deflections of a unipivot galvanometer. When a more 

 sensitive galvanometer is substituted for the unipivot 

 instrument, in a set of observations the sensitivity for 

 current was 3x10-" amperes with a sensitivity for 

 voltage of 1-2 X 10-* volts, although the valves used 

 were rather soft and not suitable for high current 

 sensitivity, and no amplification device was used. 



J. C. M. Brentako. 



Manchester University, Manchester, December 10. 



Simple Sensitive Flames. 



It does not seem to be well known that it is quite 

 easy to make a very simple flame sensitive to sounds 

 of short wave-length for working with gas at ordinary 

 supply pressure. 



Such a sensitive jet is obtained from a glass tube, of 

 diameter about i cm., simply by rotating the end of 

 the tube in a blow-pipe flame, so that it takes the 

 shape of a dome with an orifice. This shape provide? 

 the sudden change of pressure on which sensitiveness 

 depends. While the glass is still soft it is slightly 

 flattened so as to make the orifice noticeably elliptical. 

 As with high-pressure jets, the flame loses sensitive- 

 ness if the orifice be too nearly circular, or if its 

 ellipticity be too great. The best condition is attaint 

 by trial. If the orifice be too large the flame is easilj 

 disturbed by draughts or by slight variations in thi 

 pressure of the gas. If it be too small the normj 

 gas pressure is too low to give a flame on the point 

 of flaring, a well-known condition of sensitiveness. 

 These low-pressure flames are sensitive to lowe^ 

 pitched sounds than is usual with high-pressur 

 flames. Like the latter, they are most sensitive 

 sounds in a particular direction, viz., the direction 

 the major axis of the elliptical orifice. 



G. A. Sutherland. 



University College, W.C.i, December 6. 



