April 20, 1893] 



NATURE 



597 



lamp : and proves to he equal in delicacy and accuracy of test- 

 ing 10 Liveing's indicator and other forms of apparatus of 

 precision at present in use. The lamp presents the great 

 advantage of serving at once for lighting, for ordinary gas- 

 testing by the oil-flame, and for most accurate and delicate 

 testing by means of the hydrogen flame. 



The paper gives full statements of the results of the flame-cap 

 measurements of the new lamp, and of the lamps mentioned 

 above. 



The general conclusions to be drawn from these measure- 

 ments, and from experience derived from working with the 

 different lamps, are the following : — 



(i) The indications of the Pieler lamp begin at the lowest 

 limit of 025 per cent., but quickly become too great to be 

 utilised. The thread-like tip extending above the flame for 

 several inches in pure air must not be mistaken for a cap, but it is 

 scarcely distinguishable from the cap given by 0*25 per cent, 

 of gas. 



This lamp suffers under the disadvantage that much of the 

 feeble light of the caps is lost by the obstruction of the gauze : 

 the gauze also frequently presents a bright reflecting surface 

 behind the flame, and this renders the observation of the cap 

 impossible. All the other lamps in use are free from these 

 interferences due to the gauze, and if their glasses are blackened 

 behind internally by smoking them with a taper they become 

 well suited for the observation of caps. 



(2) The Ashworthbenzoline lamp begins its indications doubt- 

 fully at 0'5 per cent., the cap thus produced being more distinct, 

 i)ut not greater in height, than the mantle of the flame seen in 

 gas free air. 



But starting with certainty with an indication of i per cent., 

 it gives strikingly regular indications up to 6 percent., and 

 even higher percentages may be read off ia a lamp with a long 

 glass. 



(3) The standard 10 mm. hydrogen flame gives distinct 

 indications from 0-25 to 3 per cent. ; the cap then becomes 

 too high for measurement in the lamp ; but by reducing the 

 (lame to 5 mm., cap readings may be taken up to 6 per cent, 

 of gas. 



The lower indications may similarly be increased by raising 

 ihe flame to 15 mm. 



(4) The oil Ibme produced by unmixed colza oil gives no in- 

 dications with percentages below 2. With i per cent, of gas 

 I he flame from colza mixed with an equal volume of petroleum 

 (water- white) produces an apparent cap, which, though some- 

 what more intense than the natural mantle seen in gas-free air, 

 is only equal to this mantle in dimensions, and might easily be 

 mistaken for it. 



The oil flame, when it is reduced until it just loses its 

 luminous tip, however, gives distinct indications from 3 to 6 

 per cent. 



The largest indications are produced by drawing down the 

 flame in the presence of the gas, until a cap of maximum size is 

 obt?.ine<l. 



A carefully regulated oil flame may, therefore, conveniently 

 sapplement the hydrogen flame for the indication of gas varying 

 from 3 to 6 per cent., and in the new hydrogen lamp this will be 

 found to be a convenient method to adopt. 



The use of colza alone in the oil-lamp is very inconvenient 

 for gas-testing : the wick quickly chars and hardens on the 

 top, and cannot then be reduced without danger of extinc- 

 li m ; it can never be obtained .satisfactorily in a non-luminous 

 con<lition. The admixture with petroleum obviates these 

 difliculties. 



The use of the hytlrogen flame for gas-testing has been pro- 

 posed, but hai never been hitherto carried into practice in an 

 ordinary safety lamp. Careful comparison proves this flame to 

 be superior to the alcohol flame and to all other flames at 

 l)resent sugi^ested. Its indications have never been carefully 

 observed and measured before ; they are carefully summarised 

 in the present paper. 



It will be readily understood that the main advantages result- 

 ing from the use of the hydrogen flame are the following : — 



(1) The flame IS non luminous, whatever its dimensions may 

 be, and therefore does not interfere with the perception of 

 the cap. 



(2) The flame can always be adjusted at once to standard 

 height and maintained at that height sufficiently long for the 



NO. 1225, VOL. 47] 



completion of the test ; whereas other testing flames are con- 

 stansly varying in dimensions, and most of them cannot be set 

 to standard size at all with any certainty. 



Thus a colza-petroleum flame exposed in air containing alow 

 percentage of gas when twice adjusted gave caps of 8 and of 

 20 mm. The reduced oil flame often fell so quickly that cap- 

 readings with low percentages of gas could not be taken at all. 



(3) The caps produced over the hydrogen flame are larger 

 than those produced by any flame of corresponding size. 



(4) The size of the hydrogen flame can therefore be so far re- 

 duced as to enable it to be used in an ordinary safety-lamp. 



The size of the flame may further be suitably varied so as to 

 increase or decrease the height of the cap and thus either in- 

 crease the delicacy of the tdt or extend its range. 



(5) The hydrogen flame shows no trace of mantle or cap in 

 air free from gas ; it resembles the Pieler flame in showing only 

 a slender thread above its apex. The colza-petroleum and the 

 benzoline flames show pale mantles in gas-free air, which may 

 be easily mistaken for a small percentage of gas. 



(6) The standard hydrogen flame burns vigorously, it is of 

 fair size, and cannot be extinguished by accident ; whereas the 

 reduced flames ordinarily used in testing burn feebly and are 

 readily lost. 



(7) Hydrogen is supplied pure and of practically invariable 

 composition ; whereas oil and alcohol are apt to vary much in 

 composition, and therefore to give flames whose indications vary 

 with the sample of liquid which is being burnt. 



It should be noted that the hydrogen flame is set to standard 

 sizeiw the presence 0/ the gas, and therefore yields accurate indi- 

 cations in any atmosphere in which the test is made. 



The paper gives full descriptions of the method pursued for 

 obtaining accurate flame-cap measurements in this research. The 

 indications furnished by the new lamp in air co uaining coal gas 

 and water-gas are also tabulated ; and it is shown that these gases 

 are readily detected when present in small proportions in the 

 air, and their amount is accurately determined. The lamp 

 shows equal delicacy and accuracy in the detection and estima- 

 tion of petroleum vapour in the air. 



When used for the detection of fire-damp the amount of fine 

 coal-dust ordinarily present in the air of the mine caused no in- 

 terference with the test. The lamp had been proved by use in 

 the coal-mine to be thoroughly practical and easy in its applica- 

 tion togas-testing. 



February i6. — "Further Experiments on the Action of 

 Light on Bacillus anthracis." IV. By H. Marshall Ward, 

 D.Sc, F. R.S., Professor of Botany, Royal Indian Engineering 

 College, Coopers Hill. 



The author has continued his experiments, proving that 

 the light of a winter sun and that of the electric arc rapidly 

 destroy the life of the spores of the anthrax bacillus, and 

 showing that the bactericidal action is really direct, and not 

 due to elevation of temperature, or to any indirect poisoning or 

 starving process incident on changes in the food materials. The 

 evidence goes to prove that the effect is chiefly if not entirely 

 due to the rays of higher refrangibility in the blue-violet of the 

 spectrum. 



The experiments have been continued with special reference 

 to these latter points, and confirm the general conclusions in 

 every detail. Not only so, but the further results prove that the 

 inhibitory and deadly effects of direct insolation are not confined 

 to Bacillus anthracis, but also extend to other bacteria and 

 even to the Fungi ; and throw some light on several pro- 

 blems which have presented themselves during previous in- 

 vestigations. 



Experiments with Coloured Screens of Various Kimls. 



The author described experiments made during December to 

 February with coloured screens of various kinds ; premising that 

 the methods employed in preparing and exposing the plates, &c., 

 have been the same as those referred to in the previous com- 

 munication. 



The results show that when plates are exposed for equal 

 periods behind screens transmitting blue and violet rays, and 

 behind screens which cut off those rays, the spores on the former 

 are killed, whereas no bactericidal action occurs on the latter. 



Experiments with Spores and Food Material on Separate 



Plates. 

 In order to test still further the accuracy of previous con- 

 clusions, that the bactericidal action of the sunlight is direct, 



