214 



NATURE 



[June 28, 1900 



sion in living plants, and it is pointed out that the structure of 

 a typical herbaceous leaf illustrates in a striking manner all 

 the physical properties of a multiperforate septum. Regarded 

 from this point of view it is shown that the stomatic openings 

 and their adjuncts constitute even a more perfect piece of 

 mechanism than is required for the supply of carbon dioxide for 

 the physiological needs of the plant, and instead of expressing 

 surprise at the comparatively large amount of the gas which an 

 assimilating leaf can take in from the air, we must in future 

 rather wonder that the intake is not greater than it actually is. 



From data afforded by actual measurements of the various 

 parts of the stomatal apparatus of the sunflower it is shown that 

 an extremely small difference of tension of the carbon dioxide 

 within the leaf, as compared with that in the outer air, will 

 produce a gradient sufficient to account for the observed intake 

 during the most active assimilation. 



It is also shown that the large amounts of water-vapour which 

 pass out of the leaf by transpiration are well within the limits of 

 diffusion, and that it is unnecessary to assume anything like mass 

 movement in the outcoming vapour. 



The translocation of solid material from cell to cell in the 

 living plant is next considered, especially with reference to this 

 transference being, at any rate in part, brought about by means 

 of the minute openings in the cell-walls through which the con- 

 necting threads of protoplasm pass. Notwithstanding the very 

 small relative sectional area of these perforations they probably 

 exercise an important function in cell-to-cell diff'usion, in virtue 

 of their properties as multiperforate septa. 



There are two appendices to the paper, one in which a full 

 description is given of a series of experiments on the absorption 

 of carbon dioxide by solutions of caustic alkali from air in 

 movement ; the second being devoted to a detailed description 

 of the methods used for accurately determining the carbon dioxide 

 absorbed. 



Physical Society, June 22.— Mr. T. H. Blakesley, Vice- 

 President, in the chair. — A paper, entitled " Notes on Gas Ther- 

 mometry," by Dr. P. Chappuis, was read by Dr. Marker. The 

 author having been led to recognise that hydrogen could not be 

 used as a thermometric substance at high temperatures, on 

 account of its action on the walls of the glass reservoirs, has had 

 recourse to a constant volume nitrogen thermometer with an 

 initial pressure slightly under 800 mm. The value of the 

 coefficient of expansion of nitrogen at constant volume is variable, 

 diminishing up to 80° C. and then increasing slightly. In fact, 

 nitrogen at 100° C. behaves like hydrogen at the ordinary tem- 

 peratures, its compressibility being less than that required by 

 Boyle's law. A table of corrections was therefore prepared. 

 The readings of the constant volume nitrogen thermometer are 

 too low, but the corrections are small, amounting to about 

 0-04° C. at the temperature of boiling sulphur. The mean result 

 of the author's experiments for the boiling point of sulphur is 

 445° 2 under a pressure of 760 mm. Callendar and Griffiths' 

 results obtained with a constant pressure air thermometer is 

 444° 53. The difference is attributed to the joint action of 

 several causes: — (i) The corrections for a constant pressure 

 thermometer are about double those of a constant volume instru- 

 ment. This correction applied to Callendar and Griffiths' result 

 would raise it about o-i°. (2) Callendar and Griffiths have used 

 a value for the gas constant which is larger than that obtained 

 by more recent experiments. Adopting the latter value, the 

 boiling point would be raised to 445". (3) The divergence may 

 be due to the expansion of the reservoir. The most accurate 

 way of determining this is by the interference method of Fizeau'. 

 This method is used with small pieces of the material, and the 

 author has employed it to determine the coefficient of expansion 

 between 0° and 100°. Extrapolation to 450° might cause 

 errors The linear expansion has recently been determined by 

 Bedford between 0° and 84.0° by a comparator method. The 

 homogeneity of porcelain is doubtful, especially when glazed, 

 and the great differences occurring between the expansions 

 obtained from the above methods is attributed to the change in 

 form of the tube in Bedford's experiments, brought about by un- 

 equal thickness and want of homogeneity and consequent 

 unequal expansion. The author therefore adheres to his value 

 of the boiling point obtained from the expansion by the Fizeau 

 method, whilst recognising the uncertainty attaching to the 

 application of the coefficient of expansion of the reservoir over an 

 interval four times as great as that over which it was determined. 

 — A paper on a comparison of impure platinum thermo- 



NO, 1600, VOL. 62] 



meters, by Mr. H. M. Tory, was read by Prof. Callendar. 

 The object of this paper is to investigate the probable order 

 of accuracy attainable in the determination of high temperatures 

 by the use of ordinary commercial specimens of platinum wire. 

 Five wires were compared, from 400° to 1000° C. The funda- 

 mental coefficients of the wires varied within 40 per cent, of the 

 maximum value, but the temperatures observed by them when 

 calculated on the platinum scale by means of the ordinary 

 simple formula, did not differ by more than 9° at 1000° C. 

 Each wire was directly compared with a pure standard wire, 

 the two being wound side by side in the same tube. Curves 

 have been drawn with the platinum temperatures of the standard 

 wire as abscissae, and the differences between the temperatures 

 indicated by the two wires compared as ordinates. These curves 

 are all straight lines, within the limits of observation, and hence 

 the determination of two constants is sufficient to enable us to 

 compare an impure platinum thermometer with the standard, 

 and therefore with the scale of the gas thermometer. The two 

 constants can at once be obtained from observations at the boil- 

 ing point of sulphur and the freezing point of silver, and thus a 

 practical thermometric scale can be established, which between 

 o°and 1000° never differs by more than two or three degrees from 

 the gas scale.— Prof. Callendar said he was unable to agree with 

 the correction to his observations suggested by M. P. Chappuis. 

 He considered that the uncertainty in the coefficient of expan- 

 sion of the gas was due to uncertain changes in the volume of 

 the bulb, and to uncertainty in the coefficient of expansion of 

 mercury. The fundamental coefficient of mercury was 

 •00018153 according to Regnault, •00018216 according to the 

 later reduction of Broch, and •00018256 according to experi- 

 ments by Chappuis with a hard glass bulb. It made a difference 

 of no less than 4 per cent, in the fundamental coefficient of ex- 

 pansion of the glass, according as the original results of 

 Regnault, or the value found by Chappuis, assuming the linear 

 expansion of the glass, were adopted. The importance of the 

 changes in the volume of the bulb had been fully pointed out, 

 and a method of taking approximate account of these changes 

 had been explained in the paper on the boiling point of sulphur 

 in 1890. Unfortunately the glass employed was rather soft, and 

 the changes of volume which occurred were too great to 

 permit of the most accurate determination of the coefficient. 

 The boiling point, when corrected for the smaller expansion of 

 the bulb, came out lower than 444'53°. With regard to porce- 

 lain, Prof. Callendar did not consider it a good material, on 

 account of the glaze. He did not think that the average co- 

 efficient of a tube or bulb over a large range of temperature could 

 be inferred from a small and possibly asymmetric specimen. The 

 results might be less inconsistent in the case of homogeneous and 

 well-annealed metallic bulbs. The correction for the expansion 

 of the bulb was, he believed, given by the expression ilt = 

 {(Z-\-bQ)t{t- 100). He did not agree with M. P. Chappuis that 

 the correction w^s independent of c, although the va'ue of b 

 was certainly most important at high temperatures. He also 

 wished to take exception to the method adopted by Chappuis of 

 calculating the correction of the nitrogen thermometer. Accord- 

 ing to Joule and Thomson, the correction should be greater ; 

 according to other authorities, it might be less. He hoped to 

 discuss this in a further communication to the Society. Mr. 

 Glazebrook said that, although he placed confidence in Chap- 

 puis' formula for a definite piece of porcelain between certain 

 temperatures, he thought further and careful work was necessary 

 before fixing on a formula for ordinary use. Prof. Carhart said 

 he would like to see a comparison made between the results of 

 experiments with gas thermometers and those with platinum 

 and platinum-rhodium couples. Mr. Rose-Innes expressed his 

 interest in the behaviour of nitrogen about 100° C, as mentioned 

 in M. P. Chappuis' paper. Dr. Lehfeldt said the peculiarities of 

 the nitrogen scale between 70" and ^'0° might be explained by 

 the reversal of the properties of nitrogen between o°and 100°. — 

 A paper on the law of Cailletet and Mathias and the critical 

 density was read by Prof. S. Young. The law of Cailletet 

 and Mathias is very nearly, though in most cases not absolutely, 

 true. It appears to be only strictly true when the ratio of the 

 actual to the theoretical density at the critical point has the 

 normal value 377. The curvature of the " diameter" is gene- 

 rally smaller the nearer this ratio approaches its normal value. 

 The curvature is in nearly every case in opposite directions, 

 according as this ratio is greater or less than 377. The curva- 

 ture is generally so slight that the critical density may be calcu- 

 lated from the mean densities of liquid and saturated vapour at 



