]\gt ; Experiments with Mercury Jets .
By S. W. J. SMITH , M.A. , D.Sc .
, F.B.S. , and H. Moss , M.Sc .
, Imperial South Kensington .
( Received February 22 , 1917 .
) Section .
Rcl the 1 .
troductory.\mdash ; Certain electro-chemical experiments , to be described in a subsequent communication , arose out of peculiarities which we noticed in the behaviour of a stream of mercury issuing from the lower , drawn-out , end of a vertical tube that liquid .
The salient feature of the phenomena is the relation between the length of the continuous part of the liquid stream , called for convenience the -length , \ldquo ; and the " " head\ldquo ; under which the mercury is ejected .
The experiments were made with orifices of diameters varying from about mm. downwards to about mm. It was found that the relation between the and the head depended upon the form of the nozzle through which the mercury escaped .
For definiteness the nozzles were all constructed in the same nely by drawing out a piece of capillary tubing of about 1 mm. bore and severing it , as nearly as possible perpendicularly , at the centre of the constricted part .
The nozzle so obtained tapered , in between 1 and 2 cm .
, from the original diameter of the tubo to the diameter of the orifice and was very nearly cylindrical at the point of discharge .
The values of the jet-length were determined by means of a cathetometer microscope .
2 .
Fcatures of the for Jcts of Different Constant Surfacc \mdash ; In bhe particular experiments represented in fig. 1 , the mercury fell into a solution of mercurous nitrate ( of strength -mol .
per litre ) contained in a vessel with flat glass sides .
The object of using this solution was to ensure that the surface tension of the mercury should be ) ractically constant hout the measurements .
Considering the curve ABCDE , of the figure , which refers to an orifice of diameter mm. , it will be seen that between A and , where the heads are at first only just sufficient to cause the jet to form , the jet-length increases very rapidly as the head rises .
At the rate of increase of with changes suddenly , becoming smaller and remaining practically constant till is reached .
Beyond this point the th falls , very rapidly at first , as the head is increased .
At the higher heads ( near E ) the rate of decrease of jetlength is comparatively small .
Between and the jet is relativel .
unstable , Dr. S. W. J. Smith and Mr. H. Moss .
lengthening and shortening in a capricious way which makes definite measurements of its length difficult .
Sometimes it is impossible to obtain anything more than rough values of its mean It will be seen that the other jets give lh curves of the same character as that .
just described .
In each case there are two critical points at which the value of changes suddenly .
The critical jet-lengths decrease and the critical heads rise as the diameter of the orifice diminishes .
3 .
The Curves for of Constant Diameter and Surface Tensions.\mdash ; In order to find how the positions of the critical points depended upon the surface tension we made the observations represented in fig. 2 with an orifice of mm. diameter .
The different curves were obtained using solutions of of different strengths , it being known that the surface tension between mercury and a solution of decreases as the salt-content rises .
The points marked by circles were obtained when distilled water was used .
It will be noticed once that , although the critical jet-length does not alter greatly , the critical head rises as the surface tension increases .
4 .
Critical Vetocities of Bfflux.\mdash ; It is sufficient for the experiments referred to in S1 that the phenomena exhibited in figs. 1 and 2 exist ; but they are of some general interest and we have therefore endeavoured to verify certain impressions as to their cause .
The sudden drop in the jet-length at the higher critical point reminded us of the experiments of Osborne Reynolds dealing with the critical velocity of Expenments with Mercury Jets .
flow of a liquid through a pipe .
It seemed likely that there might be similar critical velocities for jets of the kind we used .
In order to test this impression we performed some qualitative experiments with a water jet into which , along its axis , was fed a thin column of red ink .
This jet gave lh variations resembling those of and 2 and the ink remained a thin stream along the axis until the upper critical head was reached .
Immediately after this was passed , turbulent motion became obvious .
The ink was dispersed throughout the jet .
We could not , of course , perform similar experiments with the mercury ; but , although the water jet was comparatively very wide , it seemed reasonable to suppose that the higher critical point had the same cause in both cases .
We were , therefore , led to regard the peculiar behaviour of the jets as due to the existence of critical velocities .
From this point of view it seemed likely that the lower critical velocity was the counterpart of the phenomenon observed by Allen* in connection with the motion of spheres through liquids .
5 .
Theoretical between Variables which Critical Depend.\mdash ; If , in the case of a pipe , we suppose the only variables involved to be its diameter and the viscosity and density of the }iquid , we can ( as * Phil. Mag September , 1900 .
Dr. S. W. J. Smith and Mr. H. Moss .
is well known ) deduce a relation between the critical velocity and the val.iables by the method of dimensions .
The result is where is the critical velocity .
In the case of a jet , however , we have a free surface and therefore an additional variable , the surface ension of the liquid , upon which the critical velocities may depend .
Introducing this factor into the dimensional equation we now get cc where is the surface tension and is an undetermined index .
If we accept this result we see that must be positive since , from fig. 2 , we find that increases with .
Also , since , from fig. 1 , increases as diminishes , must be ative .
Therefore the value of must lie between zero and unity .
This is as far as we can without further experimental data , unless , for example , we assume that the viscosity , if it has any influence , must tend to increase the critical velocity .
In that case we might draw the further conclusion that must be positive and therefore that cannot exceed 6 .
xperimental Relation the Critical Velocity and the Diameter of \mdash ; The velocities of the jet were not determined in the experiments of ; but in another set of measurements ( in which the mercury fell into distilled water ) the upper critical velocities were determined approximately .
The following Table contains the results:\mdash ; Measurements of this kind cannot be made very accurately because very slight disturbances are sufficient to alter considerably the head under which the jet breaks down and it is of course impossible to guarantee that the orifices are of strictly similar character .
But the experiments obviously that the critical velocity is inversely proportional to the square root of the diameter , i.e. , that 7 .
Experimental Relation between the Critical Velocity and the Surface TcJtsion.\mdash ; Another approximate estimate of can be made by means of the Expenments with Jets .
results shown in fig. 2 .
The ratio of the surface tensions of the and solutions of was measured and found to be nearly .
The upper critical velocities for these solutions were not determined at the time , but from subsequent observations it can be inferred that their ratio was not far from .
Hence , since , the value again comes into view .
8 .
al hicance of the Results.\mdash ; These results suggest that the viscosity of the liquid has little influence upon the critical velocities and that , to a first approximation at any rate , we may write If we put this in the form we see at once , without recourse to the method of S 5 , that it is dimensionally correct .
For , as is well known , both quantities have the dimensions of essure .
The first term suggests the operation of " " inertia\ldquo ; forces ; the second the opposing influence of the pressure due to surface tension .
We return to this later on .
We may , however , remark here that we should not expect the relation to be universally true .
Its simplicity must arise from the fact that , in particular cases , the " " inertia\ldquo ; forces are predominant .
There are probably other cases in which , for example when the viscosity is large , " " viscous\ldquo ; forces are no longer negligible .
9 .
Rayleigh 's Theory of Liquid Jets.\mdash ; The quantities called " " critical velocities\ldquo ; in the preceding paragraphs are not accompanied by any perceptible discontinuities in the relation between the velocity of efflux and the head ; the changes in the jet-length are the only indication of their existence .
We therefore turned our attention next to the consideration of the factors which determine the length of the jet , and have endeavoured to apply Lord Rayleigh 's theory to the results which we have obtained .
This develops the consequences of the fact , first noticed by Plateau and applied by him to the explanation of Savart 's results , that a cylindrical liquid column is unstable when its length exceeds its circumference .
It follows from this fact that any disturbance impressed upon the liquid column at the orifice will , if its wave-length exceeds the circumference of the column , increase in amplitude as it travels downwards until , finally , the sides of the column will come together and the continuity of the jet will be The time which elapses before this occurs will depend upon the initial amplitude of the disturbance and upon the rate at which that amplitude grows .
* ' Proc. Lond. Math. Soc vol. 10 , p. 4 ( 1878 ) .
Dr. S. W. J. Smith and Mr. H. Moss .
If is the initial amplitude , the value which it has to acquire before disintegration occurs , and the time required , we may suppose these connected by an equation of the form or * Whence , if we knew the value of and of for any given disturbance , we could calculate the value , , which the jet-length would have for a given velocity ( supposed constant ) of descent of the liquid .
It is easy to show by the method of dimensions that , if viscosity be neglected , we must have Lord Rayleigh has further shown that the exact result is of the form , where is the ratio of the wave-length of the disturbance to the circumference of the jet , and that has a well-defined maximum value occurring near For equal amplitudes , therefore , a disturbance of wave-length will cause disintegration more rapidly than any other .
Moreover , if we considered disturbances of unequal initial amplitudes and , for which the values of were and , the times and required by them to produce disintegration would be connected by the equation whence , for , we should have even with if we had Hence , if the times of disintegration long enough , the disturbance with the greatest would produce disintegration sooner than any other .
Without exact knowledge of the initial disturbance of the jet , we can imagine it resolved into components of different wave-lengths and amplitudes , whose effects are superposed , and can suppose that the relation between the wave-lengths and amplitudes of these components determines the conditions under which the jet breaks down .
The time of disintegration will be given by where is the initial amplitude of the disintegrating disturbance , and is the value of with which its wave-length corresponds .
Substituting for we get Experiments with Mercury Jets .
as the simplest theoretical relation between the variables with which we are concerned .
10 .
The Difference betweour Results and those of Savart in Terms of this .-For our jets , unlike those of Savart , in which the liquid ( water ) flowed through a hole , several millimetres in diameter , in the bottom of a containing vessel , the value of was not even approximately constant .
We may take this to mean that , in our case , the quantity was variable and may assume that the variation was probably confined to The experiments of which the results are collected in fig. 3 were made , primarily , to test more closely the truth of the relation suggested by the experiments of figs. 1 and 2 .
In this figure the values of ( in cm for certain cases , are plotted against the values of ( in cm .
per The latter were determined by the amounts of mercury which escaped through the nozzle in measured times , and by dividing the volumes escaping per second by the cross-sectional area of the orifice .
The error introduced by this method of calculation is not great because , in consequence of their shortness in comparison with their initial velocity , our jets ( unlike those of Savart which were very much longer ) were nearly cylindrical .
Except at the lowest yelocities they did not taper perceptibly .
The calculated velocity probably represents , within one or two per cent. , the mean velocity of the jet .
In experiments ( a ) and ( b ) the diameter of the jet was mm. ; in experiment ( c ) it was mm. In ( a ) and ( c ) the solution was ; in ( b ) it was Dr. S. W. J. Smith and Mr. H. Moss .
The curves are , of course , of the same general character as the lh curves of figs. 1 and 2 .
11 .
Application of the Th jory to Account for th Form of the lv From the final expression in S9 , assuming constant , we get From this we see that the form of the first part of the lv curve can be accounted for if we assume to be negative when is small , i.e. , if we suppose the initial amplitude of the disintegrating disturbance decreases at first as the velocity of efflux rises .
We explain the second part if the value of chancres suddenly at the first critical poinb .
Just beyond this point is seen to be roughly equal to .
The interpretation of this would be that here is practically zero .
At higher velocities it becomes appreciably negative again and remains of this sign until the second critical point is reached .
Here , in order to explain the third part of the curve , we have to assume that acquires , comparatively suddenly , a large positive value .
We could form a picture of this sequence of events by imagining that the incipient turbulence appearing at the first critical point is just sufficient , at that point , to counteract the tendency of the amplitude to decrease as the velocity rises .
The fuller turbulence appearing at the higher critical point is , on the other hand , great enough to cause a large increase in the amplitude of the initial disturbance and a corresponding decrease in the length of the jet .
12 .
Accurate tination of the Relation bjtnoeen the Form of the lv Curve and the Value of the Surface Tension.\mdash ; The experiments ( a ) and ( b ) were to be used to test whether the relation connects the critical velocities and when varies and is constant .
We chose solutions of and of KCN for our purpose because it was desirable that the surface tensions should be ( 1 ) easily comparable , ( 2 ) as different as possible , and ( 3 ) as free as possible from time effects .
A satisfactory solution giving a low value for is easily found by using ; but it is difficult to find a correspondingly satisfactory solution for which is high .
From previous experience we knew that undesirable time effects , which are generally present ( when solutions other than those of mercury salts are employed ) , might be relatively small if we used We knew , however , that , in general , there would be a time effect , in solutions * Smith and Moss , ' Phil. Mag April , 1908 , p. 484 .
Experiments with Mercury Jets .
38 ] of KCN , tending to make the surface tension decrease as the time of contact lengthened .
13 .
Difficulties due to of Surface Tension.\mdash ; When solutions of , for which the time effect is known to be very small , are employed , the second branch BC of the lh curve ( cf. fig. 1 ) is very nearly hnear .
Hence , it is presumable that the surface tension between other solution and mercury is not constant if the of its curve does not exhibit the same peculiarity .
Fig. 4 shows the curves for the solutions ( a ) and ( b ) of fig. As will be seen , the second branch of the curve is very nearly linear , while that of the KCN curve bends upwards as rises .
This difference suggests that the surface tension between the Hg and the KCN was not constant , as it have been if we had hit upon the right concentration .
By comparison with , we see that ( as was to be expected if it varied ) the surface tension probably decreased as the time of contact lengthened .
14 .
Verification of the of S7.\mdash ; The ratio of the surface tensions of the two solutions was measured by the capillary tube method and hence corresponded with comparatively long times of contact between the mercury and the solutions .
At the second critical points of the curves the times of contact are relaGively long .
We should therefore expect to get a more favourable test of the relation from the observations at these points than .
XCIII.\mdash ; A. Dr. S. W. J. Smith and Mr. H. Moss .
from those at the lower critical points , where the times of contact arc shorter .
The obselved value of was ; whence , if the suggested relatior were true , we should have .
In fig. 5 the lv curve for is repeated from fig. 3 , but that for the KCN has been obtained by the ordinates unchanged and reducing all the abscissae in the ratic 1.25 : 1 .
It will be seen that the relation holds , within the limits of experimental error , between the upper critical velocities .
the lower critical velocities , however , we see that is greater than .
This is what we should expect if , as is likely , here exceeds A repetition of the KCN measurements at the lower velocities with a solution which had stood for some time in contact with mercury gave the dotted curve shown in the figure .
It is probable that in this case the surface tensions differed less from those at the higher velocities and , as is seen , curve agrees more nearly with that for the than the corresponding part of the KCN curve obtained earlier .
The observations therefore support the view that , for both upper and lower critical velocities , the relation is true .
But they do more than this .
They suggest , if they do not fully prove , that if the surface tension had been constant throughout the measurements , in each case , the curves obtained by plotting against would have been identical .
It is therefore worth while to attempt an interpretation of this possibility .
Experiments with Mercury Jets .
15 .
A Simple Relation Suggested by the Results of S 14.\mdash ; The relation between the variables , as given by Lord theory , can be written in the form .
Then , denoting the variables for two different solutions by the suflixes 1 and 2 , respectively , we have and and if , as the results in S14 suggest , we have vays , when ) , it follows at the same that From which , if can be considered con stant , we get the simple result that the initial ance is the same fraction of the disintegrating disturbance when the velocities and are in the ratio .
16 .
More Defermination of the between the of the lv Curve the Diamof the Jet .
of elation of S15.\mdash ; The immediate object of the experiments ( a ) and ( c ) was to test whether the critical velocities are connected by the relation , when is constant .
For this purpose it would have been sufficient to re-draw the curves and ( c ) of , keeping the ordinates the same but multiplying each abscissa by in the one case and by in the other .
We made this change in the Qcissae , but , instead of keeping the ordinates the same as in fig. 3 , we diyided each by the diameter of the jet .
We could then , in addition , see at a glance whether S 15 was confirmed .
The result is shown in fig. 6 , where , therefore , the ordinates represent the values of and the abscissae those of , for the two jets .
The curves are even more nearly alike than those of fig. 5 .
It is only in the region of the upper critical velocities that they differ appreciably .
In estimating how much importance to attach to this difference , it must be remembered that small uncontrollable differences in the experimental conditions are always present .
It is impossible , for example , to construct a pair of nozzles identical in all respects except size .
Small differences in shape , e.g. in the cylindricality of the orifice , influence the properties of the jet perceptibly and we think , from experience , that the differences shown Dr. S. W. J. Smith and Mr. H. Moss .
in are more than probably due to causes of this nature .
Moreover , the results already given in S 6 , for a series of jets , are evidence that , for the upper critical velocity , the result is , in general , not far from true .
17 .
of Section I\mdash ; We have performed a few supplementary experiments with jets , of mercury and of other liquids falling in air , but before proceeding to them it will be convenient to summarize the results already obtained .
These results apply only to mercury jets of different diameters and surface tensions falling into aqueous solutions .
The density ( and viscosity ) of the jet-fluid has remained unchanged .
( i ) We have found that there are two critical velocities and that , for any pair of jets , corresponding critical velocities are connected by the relation .
( ii ) We have taken the view that the critical velocities have their origin in certain known peculiarities of fluid motion which , we suppose , cause comparatively sudden changes , at particular velocities , in the initial amplitude of the disturbance under which the jet breaks down .
( iii ) We have then found that the ratio of the initial to the final ( disintegrating ) disturbance is the same , at corresponding critical points , for all the jets examined .
( iv ) The changes occurring at such critical points are thus always of exactly the same nature .
Change of diameter or of surface tension merely alters the velocity of efflux at which each change occurs .
with Mercury Jels .
( v ) We have found further that these regularities are not confined to the critical points .
The sequence of changes of initial amplitude between these points is again always the same .
In consequence , the results can be correlated in the following simple ( vi ) For any pair of jets , the initial disturbance is the same fraction of the final disturbance for any pair of velocities whose ratio is , and , with these velocities , the jet-lengths are in the ratio Section II.\mdash ; A Comparison with Jets of other 18 .
Critical for of Liquids in Air.\mdash ; Examination of the properties of jets of different liquids falling through air should afford an opportunity of testing our previous inference with respect to the influence of density .
A comparison of mercury jets with water jets suggested itself not only because the densities of the two liquids are very different , but also because of Savart 's results for water jets of comparatively large diameter , produced in a different way .
The time at our disposal did not permit many experiments .
Fig. 7 shows one lv curve for mercury and two and for water .
It also contains one curve for methylated spirit .
The narrower nozzle ( diam. cm .
) was used for the water only .
The wider nozzle ( diam. cm .
) was used for each of the three liquids .
386 Dr. 6 .
W. J. Smith and Mr. H. Moss .
The contains also an in , nplete series of observations for water , with a nozzle of which the diameter was cm .
If the expression in S 8 is general , corresponding velocities should be in the ratio of the values of for the different jets .
We took Quincke 's values*of for the water and mercury .
The values of and for the methylated spirit were found by direct comparison with the water used .
The critical velocities and , in the cases in which they were determined , are giyen in round numbers ( from fig. 7 ) in the Table which follows:\mdash ; It will be seen from the numbers in the final columns that is not far from true for the liquids water and mercury , and also for methylated spirit when the lower critical velocity is considered .
In the latter case , however , the upper critical velocity is far too .
Instead of being in the neighbourhood of 350 cm .
per second , as it should be to agree with the results for the other liquids , its yalue is about 800 cm .
per second .
We have not investigated this discrepancy sufficiently to be able to express a final opinion as to its cause .
We found , however , that the viscosity of the methylated spirit was 43 per cent. greater than that of the water .
In contrast with this , the viscosities of water and mercury are nearly equal .
Further experiment would , no doubt , define the extent to which the critical velocity is affected by viscosity of the liquid ( cf. S8 ) .
19 .
Analysis of Jet Besults.\mdash ; It was now of interest to compare the curves of fig. 7 by the method of S 16 .
In order to examine the relation between the water curves , it would have to plot the values of against those of ; but , to permit a 'Ann . .
Phys vol. 52 , p. 1 ( 1894 ) .
Experiments with Mercury Jets .
wider comparison , we have plotted the values of against those of for each case of S 18 .
The result is shown in fig. 8 .
It will be seen that the water curves , instead of being widely separated as in fig. 7 , are now practically coincident in the region ( marked AB ) between the critical points .
The agreement between the points ( marked by ) and the points ( marked by ) is very close .
They lie , very nearly , upon a straight line passing the origin .
Except at the lower of , where the agreement is not very good , points ( marked ) lie on the same straight lin as the others .
Thus the relation outlined in S15 appears to hold ; but , between the two critical points , it can now be expressed even more simply than before .
This happens because , if we assume that the portions of the curves lying between the two critical points are on a straight line passing through the origin and if , as in S15 , we have , * It should be mentioned that , under the experimental conditions , the length of a water jet is less easy to fix than that of amercury jet .
The end of the jet is more clearly visible in the latter case than in the former .
The wetting of the glass by the water also causes trouble at the lower velocities .
388 Dr. S. W. J. Smith and Mr. H. Moss .
it follows that the quantity where is the slope of the line AB , is constant for all the The simplest interpretation of this is ( constant ) that is constant and that therefore the initial amplitude of the rating disturbance is the same fraction of the final disturbance , for all the jets , for all velocities ' between the limits specified .
shows also another interesting result .
The methylated spirit points ( marked by .
) lie almost exactly on the same line ( AB ) as the water points .
The initial amplitude is thus not only a constant fraction of the final amplitude as with water ; but , in addition , this fraction has the same value for both liquids .
Without making a quantitative comparison it can be seen at once , from fig. 8 , that , besides being variable , the initial amplitudes are much smaller fractions of the final amplitude when mercury is These differences between mercury and the other liquids must of course have a cause .
Possibly they are connected with the fact that mercury does not wet glass .
It may be , also , that the magnitude of the initial disturbance is influenced by pulsations of pressure , within the jet as each drop forms and breaks away , whose effects ( owing , for example , to differences of density ) are not of the same relative importance in each case .
In some approximate experiments with orifices of the kind referred to below ( S22 ) we found that , although the for mercury were ( as above ) than the corresponding ones for water , the lv curves for the former liquid were now , as nearly as we could decide , parts of straight lines passing through ) origin .
With such orifices it therefore appears that , as with other liquids , the initial amplitude for mercury is now a constant fraction of the final amplitude .
20 .
nericed Compajrisons between Water Jets Iercury Jets.\mdash ; If we assume the value of to be that calculable from Lord Rayleigh 's theory , viz. , we can compare the initial amplitudes quantitatively .
Thus , from the water curves , we get very nearly , and hence .
Similarly , from the mercury curve , we get at ( the lower critical velocity ) and at ( the higher critical velocity ) .
Whence we get that the initial amplitude between A and ( water and methylated spirit ) is , in round numbers , 20,000 times as great as that at ( mercury ) , while that at Gis about 8,000 times reat as that at H. Exper iments with Jet ? .
21 .
of Lord of ) has out*that some of Savart 's results can be explained on the assumption that ' the disturbances remain always of the same aracter , so ) iime of disintegration is constant This estion with respect to 's jets , that the time of is constant , is clearly true for our water jets between A and .
We have measured both and and have found that the ratio is constant .
Another consequence of the above assumption , pointed out by Lord Rayleigh , but apparently at variance with Savart 's results , is that the jetlength for a given velocity of efflux should ) proportional .
This inference is , however , again iu agl.eement with our results .
For example , when the velocity is 500 cm .
, it will be seen from fig. thaC the jetAengths are in the ratio , while If the data upon which Savart bases his conclusion ( that the is proportional to the diameter ) be examined , ib will be found that the evidence for the conclnsion is very .
It is contain ed in two series of for orifices of which the diameters were 3 and 6 mm. respectively .
vart himself remarks that the difficulty of obtaining definite nlcasurements increases with the diameter of the jet .
The becomes increasingly dependent upon extraneous disturbances until , as he points out , it is impossible to obtain even nate measurenlents when the diameter of the orifice is 9 .
Even with the narrower.jets used by him , these effects are considerable and it will be found that only one of his sets of results for the 6 mm. orifice rees reasonably with his first " " law\ldquo ; \mdash ; that the jetis proportional to the velocity of the efflux .
all his results with this orifice are considered , they furnish as little support for his second " " law\ldquo ; as for that which we have found to be true for the jets we have used .
It is interesting to compare SavarC 's results with his rrower jet ( diameter cm which were comparatively regular , with our data .
For this purpose we may take his ' fourth series , \ldquo ; S in which the jet was protected as far as possible from outside disturbances .
If , for this set , is plotted ainst v we get a line through the origin which , curiously enough , coincides almost exactly with the line AB .
relationship is shown in , where the heads of the arrows marked 'Theory of Sound , ' vol. 2 , p. 363 ( 1896 ) .
, for different velocities of efflux .
'Ann .
Chim .
Phys vol. 53 , S .
cit. , p. 368 .
VOL. XCIII . .
2 Ir Dr. S. W. J. Smith and Mr. H. Moss .
, represent Savart 's results treated in this way .
It should be remarked , however , that in this case , the diameter of the orifice , is , even approximately , mean diameter of the jet .
For equal velocities , this jet was more than 60 times as long as the narrowest jet we used and more than tinles as long as the widest .
Lord , referring to the discrepancy between Savart 's results and theory , that " " it may well be doubted whether the of ths continuous portion obeys any very simple laws , even when external disturbances are ayoided as far as possible We have therefore thought it of interest to point out that conditions can be found under which the discrepancy between theory and experiment ppears .
22 .
of the Critical elocities upon the of the iVozzle .
There is one respect in which Savart 's results differ very noticeably from ours .
They show no evidence of critical velocities .
The jet-length continues to increase after the velocity reaches the value above which , if his jets had behaved like ours , a would have ensued .
It is easy to show that this difference is due to the difference between the orifices .
We made an approximate series of observations with an orifice of about cm .
diameter , similar in character to those used by .
It was constructed by boring a hole , bevelled from the outside , in a thin steel plate .
This plate was fixed horizontally , at the end of the supply tube , in the position previously occupied in turn by each of the glass nozzles already described .
We found for water with this orifice that , as in Savart 's experiments , the jet-length continued to increase until we reached the greatest head we could apply convexliently .
The velocity of efflux was then more than three times as great as that which , with a glass nozzle of the same diameter , would have been the upper critical velocity .
23 .
The in of the Critical Velocities.\mdash ; The results just described throw light upon the cause of the phenomena we have observed .
They show that the conditions of escape of the fluid may be such that the upper critical velocity either does not exist or relatively very high and therefore inyite a comparison between one kind of orifice and other .
This comparison caIl be made , very roughly , as follows:\mdash ; When the liquid leaves the orifice of a nozzle such as we have used , assuming that enCe \ldquo ; has not arisen within the tube , the stream-lines will be practically parallel .
Owing to viscosity , however , the velocity will be greater at the centre of the jet than it is near the sides .
This difference of velocities will not be maintained after the liquid leaves the orifice , for the .
cit. , p. 370 .
Mercury Jets .
force exerted upon the moving liquid by the walls of will disappear from each element of the jet as it forms .
But viscous forces will continue to operate between the more rapidly moving central part of the element and the less rapidly moving outer parts until the differences between their velocities destroyed .
Since the velocity along the axis , the effects of ravity , must thus tend to fall after the liquid leaves the nozzle , the axial stream-lines must tend to broaden and , while this is occurring , the velocity along these stream-lines will have a radial component from the centre outwards .
It is while the liquid particles are the movements necessary at this stage of the jet 's existence that stream-line motion seems most likely to break down .
We may suppose , in fact , that excessive radial velocity is a precu1sor of turbulence and may expect the appearance of the latter to be controlled by the means at disposal for the conversion of radial kinetic energy into other .
To express this view in another way , imagine an element of the jet of length to its radius increased from .
to by the influx at its centre of a volume of liquid kinetic energy , of which the amount may be arded as radial and tending to degenerate into energy of turbulence .
The relation between and is ; that between and , the increase in the surface of the element , is .
Hence the increase of surface energy corresponding with the infiux of is .
And if exceeds , i.e. , if exceeds , the increase of surface would be more than sufficient to replace the whole of the radial kinetic energy from which turbulence might arise and so could prevent the eneration of energy which otherwise might ensue .
Such considerations indicate , perhaps more clearly than those of S 8 , why it is that , with our nozzles , there are critical velocities which depend upon the relative values of and of In the case of an orifice like that used by Savart , the region within which there are appreciable differences of , between uous moving elements of the fluid , is of restricted extent and the effects of viscosity are reduced to a minimum .
In this case the stream-lines converge as the liquid leaves the orifice .
The velocity is greatest at the sides ; but the pressure is greatest at the centre .
The hydrodynamic forces operate to remove both these differences , and at a short distance below the orifice gravity ected ) the stream-lines become parallel , and the velocity becomes uniform , across the jet .
The cause of turbulence present in the first case has disappeared .
24 .
Summary of Section II.\mdash ; The experiments of SS 18 to 23 may be Experiments with Mercury Jets .
described as , a link between the experiments of SS1 to 16 and the well-known experiments of Savart .
of the same kind as before were used , but the jet liquid was now surrounded by air .
mercury water , relation was tested and found to be nearly true .
The mercury jet had similar properties to those previously studied .
As before , between the critical velocities , increases as increases .
Water behaves differently .
Jets of diameters between and mm. were examined .
Their properties may be formulated thus : ( i ) Between the clitical points , each lv curve forms part of a straight line passing through the origin .
Hence for each jet constant , where is the time of ration , i.e. , the time taken by each element of the liquid to pass from the top to the bottom of the jet .
( ii ) The value of this constant increases with the diameter of the jet , but when is plotted the observations for all the jets fall upon the same line .
( iii ) From this it follows that has the same value for all the jets .
( iv ) The cause of disin ration is taken to be the disturbance which originates at the orifice .
( y ) Lord 's theory leads to the result that the rate of growth of this disturbance is given by where is the initial amplitude , is a constant ( S9 ) , is the time which has elapsed since the disturbance ( vi ) If this theory is applicable , the ratio of the initial to the final amplitude for each jet is given by and is therefore the same for all , since we have found to be constant .
The initial amplitude is , apparently , a constant fraction of the diameter of the jet .
( vii ) As Lord ayleigh has pointed out , some of Savart 's results suggest a simple relation of this kind , while others gest that the relation is more complicated .
( viu ) The present experiments indicate that if narrower jets had been gnetism in to Magnetic Shielding .
393 used by Savart , the simple relation would have sufficed to define their behaviour .
( ix ) The critical velocities , they exist , obey a different law when , as in Savart 's experiments , the liquid escapes } a sharp-edged hole in a horizontal plate .
critical velocities owe their existence to the shape of the nozzle we have used .
This favours turbulence\mdash ; eddies or vortices\mdash ; as indicated in the text .
Residual gnetism in Relation to Magnetic Shielding .
By ERNEST WILSON , M.Inst .
C.E. , M.I.E.E. , Siemens 1rofessor of Electrical Engineering , and J. W. NICHOLSON , M.A. , D.Sc .
, Professor of Mathematics in the of London .
Communicated Prof. Fleming , Received February The present communication records the final of an investigation , of a somewhar exhaustive character , into the various problems which were presented by the necessity for constructing a magnetic shield suitable for large spaces , and capable of giving such a of shielding that the internal field caused by the earth should not exceed the order of nitude C.G.S. unit .
The theoretical calculation of the best form of shield , and the details of its construction , were given , together with an examination of the various methods of testing the efficiencies of large shields , in the first paper .
A study of the effect of leakage through small air spaces was made at the same time , and it became apparent that not only this problem , but several others which are vital to the production of the theoretical efficiency of a shield , needed a more careful study than they had received hitherto .
The usual process adopted for removing permanent magnetism from a shell or set of shells , by the reversals of a slowly decreasing current , ceases to be efficient when the magnetisation is very small , unless special methods are introduced , and there was previously no definite indication of the degree of accuracy with which the netic induction at any point due to a coil , wound in a helix on one member of a set of spherical shells , could be either predicted or measured .
A study of these problems was made , and the results described in a second paper .
* This investigation bas agaiu been facilitated by a grant made by the Council of the Society out of the Gore Fund .
' Roy .
Soc. Proc , vol. 92 , p. 529 ( 1916 ) .
'Roy .
Soc. Proc , vol. 93 , p. 129 ( 1917 ) .
VOL. XCIII . .
2 I

