]\gt ; vailable fomparison whese foist aiscosities o are those of Zeleny , who obtained values for ranging from 960 to 1080 , but his work was done with values of X and of very different from those considered here , and the experiments of Przibram , who obtained for the values 1005 in moist air and 1160 in air partially dried by sulphuric acid .
On the Variation of the Viscosities of the Gases of the Group .
By A. O. , D.Sc .
, Assistant in the Department of Physics , University , London .
( Communicated by Prof. F. T. Trouton , F.R.S. Received May 25 , \mdash ; Read June 9 , 1910 .
) In continuation of the determinations of viscosity already recorded , * measurements of the temperature variation for the five gases of the argon group ve now been obtained .
The apparatus used was of the same type as that described in the above papers , with certain modifications to render it suitable for working at different temperatures .
Unfortunately , the method is only practicable for a somewhat limited range of temperature .
On the one hand , the temperature must be lower than that at which the vapour ressure of mercury becomes appreciable , on account of the very viscosity of the latter ; on the other hand , it must not approach the solidifying point of mercury .
At 10 C. the vapour-pressure of mercury is about 1/ 5 mm. This was to be as much as could be safely neglected , and observations were therefore taken at two temperatures only , viz. , atmospheric temperature and the temperature of steam .
However , although these limitations render observations over a large range impossible , they still allow interesting comparisons to be made of the temperature variations of the different gases .
Incidentally , redeterminations of the ratios of the viscosities of the gases to that of air have also been obtained with an entirely new apparatus .
Apparatus .
Fig. 1 represents the apparatus in its modified form .
The capillary tube was bent twice as shown in order to reduce the total length of the apparatus 'Roy .
Soc. Proc , vol. 83 , pp. 265 and 616 .
VOL. LXXXIV .
.Dr .
Rankine .
Variation with Ternperature of [ May 25 , without seriously altering the time occupied by the mercury pellet in descending .
More than 100 cm .
of capillary tubing was thus included , although the distance between the taps and was rather less than 50 cm .
Two small bulbs and of about 1 .
each were introduced to serve as receptacles for the mercury pellet during the processes of exhausting and filling the apparatus .
The total volume between the taps was about 6 steam jacket 4 cm .
in diameter covered as much as possible of the tubes between and , a few centimetres only projecting beyond the corks .
The tubes passing through the corks were about 1 mm. internal diameter , so that their capacity represented a small fraction only of the whoIe .
The steam jacket was clamped at its middle point and arranged on an axle perpendicular to its length , so that the apparatus could be inverted as desired .
A thermometer was inserted for the atmospheric temperature only , the steam temperature being calculated from the barometric height .
The auxiliary apparatus for exhausting and filling was precisely that which has been previously described , no alterations having been found necessary .
Theory of the Jfethod .
It will be seen that the ratio of the viscosities of a gas at two different temperatures is approximately equal to the ratio of the times of fall of the mercury pellet between the marks and at those temperatures .
ertain corrections , however , must be applied .
In the first place , allowance must be made for the expansion of the glass .
Meyer 's transpiration mula as applied to this method has been shown to take the form : , where radius of capillary tube , length of capillary tube , difference of pressure below and above the mercury pellet , volume between the symmetrically placed marks and time of fall of mercury between and mean free path in the gas .
Viscositie , hese quantities R\ldquo ; depend uxpansion of lassl 8 is a very small term whose variation this respect quite negligible .
We may put in the form here / is the of the pellet and A is the cross-sectional area of the tube it .
Hence ' where is a constant as regards temperature .
Let be the linear coefficient of expansion of } lass , and the temperature difference which causes the viscosity to change from to , and the time of fall from to .
Then .
since is very small .
ThaC is to say , the ratio of the viscosities is less than the ratio of the times in the ratio .
Taking and about C. , becomes , a correction which it is necessary to make .
Secondly , it is necessary to consider whether the variation in mean free path is appreciable .
In cases where the pressure remains constant , there is a considerable alteration when the temperature rises from atmospheric temperature to C. But in this case the density and not the pressure remains constant , and a calculation shows that under these circumstances the mean free path increases by only 6 per cent. over the temperature referred to .
Since the mean free path itself only the formula as a small correction , so small an alteration in its nitude is quite ible .
Consideration was also given to the question of the variation with temperature of the surface tension of mercury .
As shown in my first paper* on this method , the unsymmetrical shape assumed by the mercluy pellet in descending produces an effective reduction of pressure .
For tubes of the size used in these experiments this reduction amounts to about 3 per cent. of the whole pressure difference , when the of the nlercury pellet is , as in this case , between 4 and 5 cm .
The question is whether this capillary diminution will vary so much with temperature as to make appreciable the effect of its variation on the total driving pressure .
The data with regard to the temperature variation of the surface tension of mercury are apparently not very reliable , but for the purpose of a small correction the decrease of surface tension for the rise of temperature from to 10 C. may be taken as amounting to about 6 per cent. in the case where air is the superposed fluid .
Upon this basis the effective increase of total driving 'Boy .
Soc. Proc 1910 , , vol. 83 , p. 265 .
184 Dr. Rankine .
with of [ May 25 , pressure the said range would be equal to 3 per cent. per cent. per cent. This correction is of the same order as that of errols liable to occur in the measurements ; but it has not been applied , owing to the uncertainty of its applicability to cases where gases other than air are in contact with the mercury .
It is true that corrections of the same order are apphed , as , for instance , that due to the expansion of the glass ; but these are perfectly definite and accurately known .
It should be pointed out , however , that the values obtained for the temperatule coefficients are probably slightly lower than the correct ones owing to this cause .
This will be referred to later .
The necessity the remaining correction appeared during test experiments with air .
It was suspected that the exposed porGion of the gas enclosed in the apparatus , i.e. , that part projecting through the , rendered the highel temperature virtually lower than that estimated .
To test this point , further experiments were made with the tubes leading to the taps filled with mercury so as to exclude air .
It was then found that mean time of fall at steam temperature was increased in the atio 1 .
Since it was not practicable to exclude the gas from the exposed gion in this manner when dealing with the rare gases , the mean times of fall at the steam temperature have been increased in the above ratio for the purpose of calculating the temperature coefficients .
Method of Observation .
The atus , having been exhausted and then filled with the gas in question , was disconnected from the pump and allowed to stand until the thelmometer indicated that the temperatnre had remained steady for a considerable time .
In no case was the variation more than one or two fifths of a degree Centigrade during half-an-hour immediately preceding the commencement of an observation .
The times of four falls of the mercury pellet\mdash ; two in each direction\mdash ; were then measured .
In order to ensure that the whole apparatus was raised to the sstimated lemperature , steam was passed through the jacket for some 15 minutes before commencing the second set of observations .
The passage of steam was , of course , continued oughout .
Again four times of fall were measured , bwo in each direction .
By taking the mean in each case the ratio of the viscosities of the gas at the two temperatures could be obtained after applying the corrections mentioned above .
It was found that , as in previous determinations , non-uniformity of the tube containing the mel.cury pellet caused the times of fall in the two directions to differ slightly .
Thus the time from to was consistently 1910 .
] Viscosities of Gases of the .
185 somewhat greater than that from to .
However , by taking the mean of equal numbers of falls in the two directions , this effect can be eliminated .
Several determinations were first of all made with air , with very consistent results , the greatest variation in the calculated temperature coefficient being less than 1 per cent. The same mercury pellet was used throughout the whole final series of experiments .
This made it possible not only to determine the temperature coefficient for each gas , but also to re-determine the relative values of the viscosities of the different gases .
Experimental Resnlts .
The results of the four determinations made with air were as follows:\mdash ; The corrections for the expansion of the glass and for the end error referred to above havs been applied in calculating these ratios .
The suffixes to indicate the temperatures .
The results indicate that temperature variations of about one-third of a degree are detectable .
In the cases of the rare gases the actual observations will be give1l .
l.\mdash ; Helium.\mdash ; Times of fall , in seconds .
Temperature C. Temperature C. 125.2 150.6 149.4 Mean of all seconds .
Mean of all seconds .
approximately .
Applying the necessary corrections , 2 .
Neon.\mdash ; Times of fall , in seconds .
Temperature C. Temperature C. to .
/ ?
to to Mean of all seconds .
Meall of all seconds .
approximately .
Dr. Rankine .
Variation with of [ May 25 , Applying corrections , 3 .
Argon.\mdash ; Times of fall , in seconds .
Temperature C. Temperature C. / ?
to to to to ' 174.4 Mean of all seconds .
Mean of all seconds .
approximately .
, corrections , 4 .
Krypton.\mdash ; Times of fall , in Temperature C. Temperature 10 C. to to to to Mean of all seconds .
Mean of all seconds .
approximately .
Applying corrections , 5 .
Xenon.\mdash ; Times of fall , in seconds .
Temperature C. Temperature C. to to to 145 .
184.2 184.2 Mean of all seconds .
Mean of all seconds .
approximately .
Applying corrections , The collected results are given in Table I. 1910 .
] viscosities of Gases of tloe Argo. .
Group .
Table I. Furbher , by comparing the times of fall with that in air , the values of the ratio , shown in Table II , were obtained .
In calculating these ratios , allowance has been made for the different slipping coefficients of the various gases , as described in my previous paper .
Table II .
of Results .
In order to make exact comparison possible in those cases where previous determinations have been made , it is necessary to assume a law of variation with temperature .
It is , of course , not possible to deduce the law from the present measurements , since they only refer to two temperatures .
For purposes of comparison , however , the law itself is not of the first importance , provided that the same temperature range is dealt with .
As a first approximation we may therefore write , where and are the viscosities at temperatures and C. respectively , and is the coefficient of increase of viscosity with temperature .
The ratios given in Table I enable us to calculate for each of the gases .
The values of for the four determinations made with air are , and \mdash ; mean The values of for the rare gases are given in the second column of Table III .
It will be noticed that with the exception of the drop from 1910 .
] Viscosities of Gases of the Argon Group .
189 Rayleigh* , Sutherland 113 , , and SchultzeS 121 .
S For helium , found as compared with 70 , and for 150 as compared with 142 .
Schultze'sl values are and for helium and argon respectively , ; but this was for larger temperature range .
If Schultze 's observations for the range from atmospheric temperature to C. are used for , it becomes 75 for helium and 156 for It will be seen that in both cases the values here recorded are somewhat lower than those of and Schultze .
It is just possible that the discrepancies partly arise from the cause previously referred to , viz. , the slight increase of total driving pressure with rise in temperature , brought about by the diminution of the surface tension of the mercury .
For reasons given earlier , the magnitude of this effect was considered too uncertain to justify application .
If , however , this were done , the value of would be increased by about 1 per cent. and that of about 3 per cent. This would bring up the values of the latter according to the present measurement to 72 in the case of helium and 146 for argon .
This still leaves a considerable divergence between Schultze 's values and the present ones ; but if the comparative difficulties attending the measurement of temperature coeflicients over small temperature are taken into consideration , this is not surprising .
Similar differences in the value of obtained by independent observers usually exhibit themselves , as , for example , the variety of values in the case of air quoted above .
to a consideration of the relative values of for the rare gases , it is surprising to find that it is higher for helium than for neon .
to Sutherland the introduction of is rendered necessary because gases are not perfect .
For a perfect gas would be zero and the viscosity would vary as the square root of the absolute temperature .
Now it is certain that the critical and liquefying points of helium are lower than those of neon ; hence , it would be natural to suppose that would be nearer thau .
This , apparently , is not the case , and it seems that in the temperature range considered neon is to be regarded as the nearest approach to a perfect gas known .
In the case of the other ases we find what would be expected , namely , that is high or low according as the critical and points are high and low .
In this connection it is interesting to note that a simple 'Eoy .
Soc. Proc , 1900 , vol. 67 , p. 137 .
' Phil. Mag 1893 , vol. 36 , p. 507 .
'Wied .
Ann 1899 , vol. 67 , p. 813 .
S 'Ann .
der Physik , ' 1901 , vol. 5 , p. 140 .
' Roy .
Soc. Proc , ] , vol. 67 , p. 137 .
'Ann .
der Physik , ' 1901 , vol. 6 , p. 302 , and 1901 , vol. 5 , p. 140 .
dimensions ocosities oonnected wheIt ixpected tdepends o : critical temperature ; and should the above simple relabion be found to be practically universal , considerable additional ) portance attaches to 1 Sutherland 's constant .
I have also made use of the temperature coefficients to -determine the ratio for comparison with the results recorded in my last paper .
The values given in Table II are for higher temperatures than the earlier determinations .
Upon reduction to the same temperatures we obtain results recorded below:\mdash ; Table The agreement is remarkably close , especially view of the fact that the aratus used in the two cases was entirely different .
This suggests that , at any rate as regards comparative measurements , the method is thoroughly reliable .
values of the viscosity for the ases are given in absolute C.G.S. units in Table VII .
These are based upon the absolute values for air reviously obtained by the author .
Gas .
Iemperature .
oyionpunt .
AirXen .

