]\gt ; On the Theory of the Tube .
By , O.M. , F.B.S. Received October 26 , 1915 ) .
A recent paper by Richards and Coombs*discusses in some detail the determination of surface tension by the rise of the liquid in capillary tubes , and reflects mildly upon the inadequate assistance afforded by mathematics .
It is true that no complete analytical solution of the problem can be obtained , even when the tube is accurately cylindrical .
We may have recourse to graphical constructions , or to numerical calculations by the method of who took an example from this very problem .
But for experimental purposes all that is really needed is a sufficiently approximate treatment of the two extreme cases of a narrow and of a wide tube .
The former question was successfully attacked by Poissorl , whose final formula [ ( 18 ) below ] would meet all ordinary requirements .
Unfortunately doubts have been thrown upon the correctness of Poisson 's results , especially by Mathieu , who rejects them ether in the only case of much importance , i.e. when the liquid wets the walls of the tube\mdash ; a matter which will be further considered later on .
Mathieualso reproaches Poissou 's investigation as implying two different values of , of which the second is really only an improvement upon the first , from a further approximation .
It must ) admitted , however , that the problem is a delicate one , and that Poisson 's explanation at a -ical point leaves to be desired .
In the inyestigation which follows I hope to have succeeded iu the approximation a stage beyond that reached by Poisson .
In the theory of narrow tubes the lower level from which the of the meniscus is reckoned is the free plane level .
In experiment , the lower level is usually that of the liquid in a wide tube connected below with the narrow one , and the question arises how wide this tube needs to be in order that the inner part of the iscus may be nearly enough plane .
Careful experiments by Richards and Coombs led to the conclusion that in the case of water the diameter of the wide tube should exceed : mm. , and that probably 38 mm. suffices .
Such smaller diameters as are often employed 20 mm. ) very appreciable error .
Here , again , we should naturally look to mathematics to supply the desired information .
The case of a straight 'Journ .
Amer .
Chem. Soc No. 7 , July , 1915 .
' Math. Ann vol. 46 , p. 175 ( 1895 ) .
'Theorie de la Capillarite , ' Paris , 1883 , pp. 46-49 .
On Theory of the wall , making the problem two-dimensional , is easy , that of the circular wall is much more complicated .
Some drawings ( from theory ) given by Kelvin , indicate clearly that diameters of 1.8 cm .
and cm .
are quite inadequate .
I have attempted below an analytical solution , based upon the assumption that the necessaly diameter is large , as it will be , if the prescribed error at the axis is small enough .
Although this assumption is scarcely justified in practice , the calculation indicates that a diameter of cm .
may not be too As Richards and Coombs remark , the observed curvature of the lower of the meniscus may be used as a test .
Theory shows that there should be no sensible departure from htness over a of about 1 The For the surface of liquid standing in a vertical tube of circular section , we have , ( 1 ) in which is the vertical co-ordinate measured ) from the free plane level , is the horizontal co-ordinate measured from the axis , is the the at any point makes with the horizontal , and where is the surface-tension , the acceleration of gravity , and the density of the fluid .
The equation expresses the equilibrium of the cylinder of liquid of radius .
At the wall , where assunles a given value , and ( 1 ) becomes .
( 2 ) If the radius ( r ) of the tube is small the total curvature is nearly constant , that is , the surface is nearly spherical .
We take , ( 3 ) where is the of the centre and the radins of the sphere , while represents the correction required for a closer approximation .
If we omit ?
/ altogether , ( 2 ) gives .
( 4 ) * Compare 'Phil .
Mag vol. 34 , p. 309 , Appendix , 1892 ; ScieDtific Papers , ' vol. 4 , p. 13 .
The reference is given below .
It may be remarked that is sometimes taken to denote the double of the above quantity .
Lord Rayleigh .
Also , if be the height at the lowest point of the meniscus , the quantity directly measured in experiment , .
( 5 ) In this approximation , and thus in terms of .
( 6 ) When the ] of contact ( i ) is zero , , and , ( 7 ) the well known formula .
When we include , it becomes a question whether we should retain the value of , i.e. , sppropriate when the surface is supposed to be exactly spherical .
It appears , however , to be desirable , if not necessary , to leave the precise value of open .
the value of from ( 3 ) in ( 1 ) , we get , with neglect of .
( 8 ) For the purposes of the next approxinlation we may omit and the integra ] , which is to be divided by .
Thus , ( 9 ) and on integration .
( 10 ) We suppose with Poisson and Mathieu that so that , ( 12 ) corresponding to ( 13 ) To determine we have the boundary condition , ( 14 ) which gives in ternls of and .
Explicitly .
( 15 ) These latter equations are given by Mathieu .
On the Theory of the Tube .
We have now to find the value of to the corresponding approximation .
the observed of the meniscus ; ( 16 ) In the important case where , the liquid wetting the walls of the simply , and .
( 18 ) This is the formula given long since by Poisson , only difference being that his is the double of the quantity here so denoted .
It is remarkable that Mathieu rejects the above equations as applicable to the case , on the ground ) then in ( 13 ) becomes infinite when .
But , with which comes into comparison , is infinite at the same time ; and , in fact , both and in equation ( 8 ) when .
It is this circumstance which really determines the choice of in ( 11 ) .
We may now proceed to a yet closer oximation , introducing approximate values of the terms previously olected aether .
From ( 13 ) and from ( 12 ) ' Nouvelle Theory de l'Action Capillaire , ' 1831 , p. 112 .
Lord Rayleigh .
Thus constant .
We have now to choose , or rather , and it may appear at first sighb as though we might take it almost at pleasure .
But this is not the case , at any if we wish our results to be applicable when .
For this purpose it is necessary that be a small quantit ) , and only a articular choice of will make it so .
For when terms vanishing when We must therefore take lltaking .
( 21 ) It should be noticed bhat so determined does not become infinite when and : For we have Also with the general value of ( 22 ) As before On the Theory of the Capillary Tube .
and .
( 23 ) The integral in ( 23 ) can be expressed .
We find .
( 24 ) The expression for in terms of is complicated , and so is the relation between and demanded by the boundary condition ( 25 ) But in the particular case of greatest interest much simplification ensues .
It follows easily from ( 25 ) that When we introduce this condition into ( 24 ) , we get , ( 26 ) and accordingly .
( 27 ) Hence by successive approximations .
( 28 ) If the ratio of to is at all such as should be employed in experiment , this formula will yield , viz. , , with abundant accuracy .
Our equations give for the whole height of the meniscus in the case .
( 29 ) Another method of calculating the correction for a small tube , originating apparently with Hagen and Desains , is to assume an elliptical form of surface in place of the circular , the minor axis of the ellipse being vertical .
In any VOL. XCII.\mdash ; A. Lord Rayleigh .
case this should allow of a closer approximation , and drawings made for by Prof. Perry suggest that the representation is really a good one .
If the semi-axis minor of the ellipse be , the curvature at the end of this axis is , and in our previous notation .
Also , being equal to and .
This yields a quadratic in ; hence approximately .
It will be seen that this differs but little numerically from ( 28 ) , which , however , professes to be the accurate result so far as the termin inclusive .
The Wide Tube .
The equation of the second order for the surface of the liquid , assumed to be of revolution about the axis of , is well known and may be derived from ( 1 ) by differentiation .
It is ( 32 ) If be small , ( 32 ) becomes approximately In the interior part of the surface under consideration may be neglected , and the approximate solution is .
, denoting , as usual , the Bessel 's , or rather Fourier 's , function of zero order , and being the elevation at the axis above the free absolutely plane level .
For the present purpose is to be so small as to be igible in experiment , and the question is how large must be .
When is small enough , may be large while still remains small .
Eventually increases so that the formula fails .
But when is large enough this occurs , we may if necessary carry on with the twodimensional solution properly adjusted to as will be further explained 'Proc .
Roy .
Inst 1886 ; 'Popular Lectures and Addresses , ' I , p. On the Theory of the Capillary Tube .
IQ the meantime it will be convenient to give some numerical examples in .
In the usual notation the values of , up to , are tabulatsd .
* In the case of water cm .
If we take , and have , so that is still fairly small .
Here for water cm .
and cm .
A diameter of cm .
is thus quite unless an error exceeding cm .
be admissible .
suppose , and take .
Then , again small .
For water cm .
, and cm .
This last value of is about given by Richards and Coombs as the maximum admissible of reading , and we may conclude that a diameter of cm .
is quite to take advantage of this degree of refinement .
We may go further in this example without too great a loss of accuracy .
Retaining , let us make .
I find , about , that the extreme value of is , still moderately small .
Here cm .
, which is thus shown to be inadequate in the case of water .
But apart from the question of the necessary diameter of tube , information for experimental purposes can be derived in another manner .
The value of ( on the axis ) is ; and when ) .
when .
For the best work should be on the limit of what can be and then and could just be distinguished .
The observer may be satisfied if no difference of level can be seen over the range in the case of water this range is cm .
, or say 1 cm .
It has already been remarked that when is small enough may become great within the limits of application of ( 35 ) .
To shorten our expressions we will take temporarily as the unit of length .
Then when is very great , Thus if be the angle thoe tangent to the curve makes with the horizontal , ; ( 37 ) equation which may be employed when is so small that a large is nsistent with a small In order to follow the curve further , up to , we may employ two-dimensional solution , the assumption being that the region of Bril .
Assoc. Rep. for 1889 ' ; or Gray and Mathews ' ' Bessel 's Functions , ' Table VL Lord Rayleigh .
moderate occupies a range of small in comparison with its actual value , i.e. a value not much less than , the radius of the tube .
On accoUnt of the magnitude of we have only the one curvature to deal with .
For this curvature so that since when is exceedingly small .
Accordingly .
( 39 ) Also ) , and .
( 40 ) The constant is determined by the consideration that at the wall thus , ( 41 ) since is small .
The value of is supposed to be the same here as in ( 37 ) , so that , ( 42 ) whence on elimination of and restoration of .
( 43 ) With sufficient approximation , when is small enough , we may here substitute for , and thus .
( 44 ) This formula should give the relation between and when is small enough , but it is onJy roughly applicable to the case of greatest interest , where , corresponding to the accuracy of reading found by Richards and Coombs .
In this case For this value of we should have .
It is true that according to ( 44 ) will be somewhat greater , but on the other hand the proper value of ( replaced by r ) is less than .
We may fairly take making cm .
cm .
This calculation indicates that a diameter greater even than those conOn the Theory of the Tube .
templated by Richards and Coombs may be necessary to reduce to negligibility , but it must be admitted that it is too rough to pire great oonfidence in the close accuracy of the final number .
Probably it would be feasible to continue the approximation , employing an approximate value for the second curvature in place of neglecting it altogether .
But although the integration can be effected , the work is rather long .
[ Added November 17.\mdash ; Since this paper was communicated , I have been surprised to find that the problem of the last paragraphs was treated long ago by Laplace in the 'Mecanique a similar method , and with a result equivalent to that ( 44 ) arrived at above for the relation between the radius of a wide tube and the small elevation at the axis .
Laplace uses the definite integral expression for , and obtains the approximate form appropriate to large arguments .
In view of Laplace 's result , I have been tempted to carry the approximation further , as gested already .
In the previous notation , the differential equation of the surface may be written In the first approximation , where the second curvature on the left is omitted , we get being the elevation at the axis , where .
For the present purpose is to be regarded as exceedingly small , so that we may take at this stage , as in ( 39 ) , S .
( 46 ) We now introduce an approximate value for the second curvature in ( 45 ) , writing , where is the radius of the tube , and making , according .
( 47 ) On integration C-cos , ( 48 ) on substitution in the small term of the approximate value of .
When is very small , so that , and , ( 49 ) is the second approximation to * Supplement au Livre , pp. 60-64 , 1805 .
Lord Rayleigh .
From We are now in a position to find by the relation the constant of integration being determined by the correspondence of .
Thus , ( 52 ) giving when is small , ( 53 ) where .
( 55 ) The other equation , derived from the flat part of the surface , is in which is regarded as large ; or .
( 57 ) In equations and are to be identified .
On elimination of , ( 58 ) in which we may put , ( 59 ) in which , since is nearly equal to may usually be neglected .
Also , in view of the smallness of and , it is scarcely necessary to retain the denominator , so that we may write .
( 60 ) On the Theory of the Tube .
The effect of the second approximation is the introduction of the second term on the right of ( 60 ) .
To take an example , let us suppose as before that , so that .
By successive approximation we find from ( 60 ) , ( 61 ) so that if cm .
( as for water ) , cm .
( 62 ) The correction to Laplace 's formula is here unimportant .
The above is the diameter of tube required to render ible according to the standard adopted .
It may sometimes be convenient to invert the calculation , and deduce the value of from the diameter of the tube ( not much less than 4 cm .
) and an approximate value of .
For this purpose we may use ( 60 ) , or preferably ( 69 ) , taking for instance .
The calculated value of would then be used as a correction .
The accompanying small Table may be useful for this purpose .
We have supposed throughout that the liquid surface is symmetrical about the axis , as happens when the section of the containing tube is circular .
It may be worth remarking that without any restriction to symmetry the differential equation of the nearly flat parts of a large surface may be taken to be , ( 63 ) so that may be expressed by the series ( 64 ) denoting the usual polar co-ordinates in the horizontal plane .
]

