]\gt ; The Pressure Distribution on the of Shelt Moving at Hi.qh Velocities .
By L. BAIRSTOW , F.R.S. , H. FOWLER , and D. .
HARTREE .
( Received December 12 , 1919 .
) S1 .
Introduction.\mdash ; The pressure at any point of a body moving through a fluid may be considered to consist of two static pressure , which would be the pressure at that point , if there were no relative motion between the body and the fluid\mdash ; and the dynamic pressure due to the relative motion .
The sum may be called the total pressure at the point , and is , of course , always positive ; the dynamic pressure alone may be negative .
Experiments , of an introductory kind , were undertaken by the authors , at the request of the Ordnance Committee , to attempt to determine , as a function of the velocity of the shell , the dynamic pressure at various points on the head of a shell , whose axis of symmetry and direction of motion coincide or nearly coincide .
The results appear to be not withoub general scientific interest , and are therefore presented in this paper by permission of the Ordnance Committee .
The authors ' best thanks are due to the officers of the Experimental Department , H.M.S. " " Excellent on whose range the firing trial was carried out ; to the other officers of the Munitions Inventions Department , who shared the work of observation and computation ; to the Director of the National Physical Laboratory , where the comparative observations at low velocities were .
made in a wind channel ; and , above all , to Mr. W. J. Goudie , of University College , London , on whose determinations of the rates of burning of powder-train time-fuzes at high rotational speeds the whole experiment was based .
The fundamental idea of the experiments is the use of a service time-fuze as a manometer to determine the pressure under which the powder is burning .
* The rate of burning of a service time-fuze is known from laboratory experiments to be affected by the total external pressure on the vents .
Let us assume the moment that this is the only factor that influences the rate of burning .
A series of observations , for shells fired along the same *A service time-fuze contains a train of gunpowder , which is ignited by a detonator pellet on the shock of discharge from the gun .
The " " setting\ldquo ; of the fuse can be varied so that a length of powder train depending on the setting is burnt before the magazine of the fuse is ignited and the shell exploded .
The setting is specified by a number which defines the length of composition burnt on an arbitrary scale .
The time of burning is taken as equal to the time interval between the iiring of the gun and the bursting of the shell .
Pressure Distribution on Shetl Moving at High Velocities .
203 trajectory at short intervals of time , determines a relation between time of burning and length of fuse burnt .
By numerical differentiation of the observations , we can obtain a relation between the rate of burning and the time , and therefore , by comparison with the laboratory experiments , between the axternat pressure the .
If the circumstances of the shell 's motion are known , this leads at once to a relation between the dynamic pressure on the fuse vents and the time , and so to the required relation between the dynamic pressure , the velocity of the shell , and the prevailing atmospheric conditions .
The external total pressure is not the only factor that affects the rate of burning .
Under identical physical conditions , the rate of burning varies from batch to batch of powder ; it is affected by the spin of the shell , and is sometimes appreciably dependent on the temperature of the fuse mixture .
In spite of these effects , owever , it became increasingly clear , as experience of time-fuzes increased , that , with suitably designed control experiments to determine the rate of burning of the fuse in the proper physical conditions , a time-fuze could be used as a manometer to determine rough quantitative of the external pressure during the flight of a shell .
S2 .
Scheme of an Experiment Based on the Fuse Jlanometer.\mdash ; In the application described in this paper , shells were fitted with caps entirely enclosing the fuse ; each cap had a series of holes drilled in it at the same distance from the nose .
The external pressure on the vent holes of the fuse is the pressure the cap ; the latter pressure is practically equal to that immediately outside the hole ( see S8 ) .
Proceeding as described in S1 , we can obtain a relation between the pressure inside the cap and the velocity of the shell along a given trajectory , for a given position of the ring of holes .
On the assumption that the presence of a hole has an } ) ortant effec we can determine the pressure-velocity relation for a series of points on the head of a shell by using caps with rings of holes at a series of positions , and draw curves ( fig. 2 ) showing the variation with velocity of the dynamic pressure at a given point .
For any given velocity we can draw curves ( fig. 3 ) showing the distribution of dynamic pressure over the head of the shell .
By integrating these dynamic pressures , we can deduce the contribution to the drag due to the dynamic pressure on the head of the shell .
The * In order to make use of these relations we must assume that the total pressure on outsi.de of the ( small ) hole the same as the total pressure at the same point of the shell in the absence of any hole .
This is the assumption made in all aerodynamical experiments on pressure distribution at low velocities .
Its application to high velocities is at present incapable of justification , either experimental or theoretical .
We shall return to this point in S8 .
Pressure Distribution on a Shell Moving High Velocities .
205 total drag must be made up of the contributions of ( 1 ) the dynamic pressure on the head , ( 2 ) the ( negative ) dynamic pressure on the base , and ( 3 ) skin friction .
If we could simultaneously determine the total drag and the two contributions of the dynamic pressure over a of velocities , we could probably make some advance to an understanding of the nlechanism of the resistance of a compressible fluid at high velocities .
It seems probable , considerations , that the pressure on the exact nose of a shell will obey closely the for1nula iven by Lord Rayleigh .
* Caps with a hole at the nose precisely were not used in this trial ; they could , however , be so used , and a direct test of Iiayleigh 's formla could be made , which would be of great interest .
The curves shown in are of the same general shape as Rayleigh 's curve for the point of the nose , but the pressure falls off so rapidly near the nose that a precise comparison on present results is not possible .
A number of difficulties , many of which will occur to the reader , have been tacitly ignored in the foregoing sections .
These will be discussed later on in S 8 .
S3 .
Details of Firing Trials.\mdash ; An 18-pdr .
field gun ( on a -angle was selected for the trial ; the muzzle velocity , determined during the experiment , was 1603 .
The ells used shown in fig. 1 .
The caps wele rounded off at the apex , and the term " " virtual point\ldquo ; will be used to denote the position of the point of a cap with a perfectly sharp point .
There were five series of caps , each with a ring of six holes .
Four series had holes of diameter inch and one series inch ; these series will be referred to by letters as Table I. point osured a ' Series was inserted as a control , to determine whether the effect of the size of the hole on the pressure was likely to be considerable .
" " Aerial Plane Waves of Finite Amplitude ' Scientific Papers , ' vol. 5 , p. 610 .
208 Messrs. L. Bairstow , R. H. Fowler , and D. R. Hartree .
Groups of five rounds were fired at fuse ettings 3 , 6 , 9 , 12 , 16 , at an elevation of in each series .
The extra rounds in series were fired at .
The majority of the rounds in series were fired by day , and the position of the shell bursts was observed .
The remaining rounds of all series were fired by night .
Times of burning of the fuzes were recorded for all rounds fired ; the means for each.roup niven in Table II .
Table II.\mdash ; Mean Observed Time of urning for Group , and Average Besiduals*of the Rounds in each Group , with the Recalculated Times of Burning .
Times in Zero point taken as fuse settings .
( See S4 .
) The following abnormal ounds are omitted from means and average residuals ( see S8):\mdash ; Series : Set 20 , ; Set 16 , ; Set 12 , , Set 12 , ; Set 3 , , Set 16 , ; Set 12 , Series : Set 12 , ; Set 9 , ; Set 6 , Series : Set 6 , ; Set 3 , The times were recorded to the nearest second , but the time for any one round is not accurate to this degree .
Experience has shown that under favourable conditions , such as obtained in this experiment , and with three or The arithmetic average difference of the members of the group from the arithmetic mean of the group .
Pressure Distribution on Shell at High Velocities .
209 more observers , the time for any round can be determined with a probable error of second .
By means of the observations of position , the ballistics of the shells were determined .
The conditions in the upper air were known from aeroplane observations .
The wind structure was also observed .
From these observations and routine calculations which need not be specified , the following quantities were determined and tabulated for the position of the shell at every second after gunfire along both the and tra , jectories:\mdash ; ( 1 ) The relative velocity of the shell and the air , ( 2 ) The static pressure of the air , * ( 3 ) The density of the air , ( 4 ) The ratio , where is the velocity of sound in the air .
S4 .
Det of the Bate of of the \mdash ; In orcler to determine the rate of burning of the fUzes used under the physical conditions , it is essential that laboratory experiments should be carried out with fuzes filled with the same batch of powder and under the same filling conditions as the fuzes used in the gun trials .
Fuzes fulfilling these conditions may be described as 's of the lot .
The rate of burning of a fuse of given lot at given pressure and temperature varies largely with the spin , especially at low pressures .
It is therefore essential to out the laboratory experiments with the fuse spinning at the correct rate , and sonl e times ) at the correct temperature .
As the velocity of the shell cannot affect the fuse , and the deceleration is small compared to the radial acceleration in the powder channels , the physical conditions of an actual firing are then properly represented in the laboratory experiment .
The initial spin of the shells in the trial was 11,660 revolutions per minute , and it may be assumed that it remains roughly constant for the fairly short times of flight with which we are concerned .
In Table III the rates of burning of the fuzes used are given , determined at a spin of 11,750 revolutious per minute .
This determination was effected by Mr. Goudie by coupling a dummy fuse body , holding the rings the powder train , to the shaft of a de Laval turbine ; the dummy was run in an air-tight box with an automatic control of the pressure .
Experiments were also carried out to determine the effect of temperature and change of spin .
Both these effects were negligible for the present trial .
The arbitrary graduations defining the fuse setting on these fuzes run from to 22 , and the length of powder burnt varies linearly with the fuse setting on this arbitrary scale .
Consequently , under constant external conditions , the time of burning is a linear function of the graduations or fuse 210 Messrs. L. Bairstow , R. H. Fowler , and D. R. Hartree .
Table III.\mdash ; Showing the Relation between the Rate of Burning and the Pressure on the Vent Holes for the Fuzes used in this Experiment , determined by Mr. W. J. Goudie at University College , London .
Spin.\mdash ; 11,750 revolutions per minute .
The rate of burning is given in fuse settings per second .
The pressure is given in atmospheres .
setting except for very short times .
The rates of of Table III are given in settings per second , and apply to all stages of the combustion under constant external conditions .
In order to construct from the observations , as accurately as possible , a curve of fuse ainst time of burning ( see fig. 4 , S8 ) , it is important to determine the constant in the above linear relation .
This constant may be regarded as the fuse setting which burns for zero time and is called the zero point of the fuse .
It is not necessarily at the zero graduation .
Its position depends on the internal arrangement of the powder-train connections , and it must be determined by experiments on the actual fuse bodies used .
It appeals that the zero point of the fuzes used was settings .
It is assumed that this zero point remains unaltered when a fuse is fired from the gun .
If this assumption is incorrect , the rates of burning observed during firing trials at short times will be in error , but those at later times will be unaffected .
S5 .
Analysis of the Observations.\mdash ; The data provided by the observations described in SS3 , 4 , are sufficient , on analysis , to provide information as to the value of the dynamic pressure for values of between and for *Note added , March 10.\mdash ; Owing to the effect of an uncertainty in the zero point ( see S8 ) , the values of the dynamic pressures for values of greater than must be given less weight than those for values of and less .
Pressure bution on a Moving at High Velocities .
211 each position of the holes in the cap. It is not neaessary to describe in detail the numerical methods used .
After the final results for the dynamic pressures were obtained , the inverse process was gone through of integrating the corresponding rates of , to obtain the time for each fuse setting as a check on the accuracy of the work .
These recalculated times are given in Table II , and it will be seen that the agreement with the observed time is satisfactory for all series , the difference between the two for any one group being less than half the average residual for the observed group in nearly all cases .
In series , which was fired at and , the recalculated times of barning are systematically slightly too short at and too long at at long settings .
The differences , however , are not than those for other series .
In the usual aerodynamical notation , the dynamic pressure may be put in the form where is the velocity , is the air density , is the velocity of sound , and is a numerical coefficient known as the pressure coefficient .
This coefficient may ( by the theory of dimensions ) depend on other arguments besides , but it is unlikely that any others are of serious importance in this connection .
It is known that is roughly constant at low velocities ; this constancy probably persists nearly up to .
If pressure , density , and velocity are measured in any self-consistent set of units , is independent of the set of units chosen , and is more suitable than the dynamic pressure itself for tabulation as it does not depend on .
The values of , determined by these trials were calculated from the dynamic pressures , and are ta bulated in Table as functions of .
They are also shown graphically in fig. 2 .
Values of the pressure coefficients , determined in a wind channel at the National Physical Laboratory , at a velocity of 40 , are also given in Table for the same } ) ositions of the holes along the axis of the shell .
The dynamic pressures at this velocity were determined more exhaustively than this along the shell .
It will be seen in Table or fig. 2 , that the observed values of fit on satisfactorily to the wind channel values .
By plotting the values of in Table for given values of , opposite the position of the holes at which was determined , a diagram is obtained , fig 3 , showing the distribution of amic pressure ( coefficients ) along the shell at high velocities .
The curves are drawn for values of equal to , and ( wind channel velocity ) .
The values of the 214 Messrs. L. Bairstow , R. H. Fowler , and D. R. Hartree .
In fig. 3 the exact shape of the curve the nose and the of the holes in series A is naturally not very definite .
This is not important for the purpose of S6 , but it could be wished that thecurves were determined by observation behind the position of the holes in series D. S6 .
Integration of the Pressure over the Head of the Shell.\mdash ; It can easily be shown that if is the dynamic pressure at the point on the shell at which the distance from the axis is , then the total force due to the dynamic pressure on the shell in its symmetrical position is along the axis and is given by the expression where is the radius of the shell .
The mean component of the dynamic pressure on the head , parallel to the axis of the shell , is therefore and the coefficient of this contribution to the } , or mean dynamic pressure coefficient , satisfies the equation This coefficient has been calculated for several values of , and the results are given in Table and plotted in fig. 2 .
Again , if we express the total drag on the shell in the form , ( 2 ) where A is the area of the cross-section of ths shell , then the drag per unit area is given by .
The coefficient is directly comparable to , the coefficient of the mean dynamic pressure over the head , which latter is one of its main contributors .
The values of , the coefficient of the drag per unit area , are not known really accurately for shells of this shape .
They certainly lie , however , between the two sets of numbers given in Table and the two curves of fig. 2 , approximating to the lower values at high velocities and the higher values at low velocities .
The pressure coefficient , corresponding to a total vacuum over the base , is clearly given by the equation , ( 3 ) where is the ratio of the specific heats of the gas .
Values of are Pressure Distribution on a Shell Moving at High .
215 also tabulated in Table and plotted in fig. 2 .
It will be seen that the coefficient is less at all velocities here reached than the coefficient alone .
The vacuum on the base is therefore by no means complete at a velocity represented by It would be of very great interest if the precise degree of vacuum could be properly determined by experiment .
This would not be possible by the use of a powder train fuse , for no such fuse will burn at all at the necessary low pressures .
S7 .
An immediate value attaches to the results now recorded , as they provide an adequate explanation of the complex behaviour of fuzes .
The changes in rate of burning which accompany change of muzzle velocity or change of diameter of shell are all explained qualitatively by curves of the kind shown in fig. 2 .
It is now possible to predict the behaviour of fuzes before trial in a manner , and with a degree of certainty , unattamable prior to the war .
It is to discuss briefly the records of pressure-distribution on the head of a shell in connection with other knowledge bearing on the point .
The curves of fig. 2 show that at all speeds in the range covered the pressure has a maximum positive value at the nose of the shell .
The pressure falls rapidly as the point of observation moves towards the base , and is negative some distance before the cylindrical part of the shell is reached .
The determination of the pressure at the nose has only been made at low speeds , but is then found to be independent of- the viscosity of the fluid medium .
At these low velocities , calculations and experiments have shown that this independence of viscosity extends so far as sensibly to cover all positive dynamic pressures .
The maximum observed pressure at the nose agrees with that calculated on the basis of an inviscid fluid , and the degradation of energy caused by viscosity shows an increasing effect as the point of observation recedes from the nose .
The complete recovery of pressure on the base of a shell which is indicated by the theory of an inviscid fluid is not even approximately represented by the experimental value .
The reduction of the observations of the trial shoots has been made on the hypothesis that the effects of viscosity on the pressure on the head of a shell are negligible , and the results are quite consistent with that hypothesis .
It may be expeoted , therefore , that the pressure could be calculated approximately from the equations of motion of a body in an inviscid compressible fluid .
In the event of a solution of those equations , the observations now recorded would provide a suitable check on the soundness of the initial assumptions .
Viewed as an addition to the experimental investigation of the motion of 216 Messrs. L. Bairstow , R. H. Fowler , and D. R. Hartree .
shells , the trials have an importance which is too obvious to require detailed notice .
S8 .
Possible Disturbing Factors , and Accuracy of the Results.\mdash ; In conclusion it may be well to return to a number of difficulties which have been more or less ignored in the earlier sections .
We have stated that the pressure at the outside of the holes in the cap is practically the same as the pressure on the vent-holes of the fuse inside .
The factors that might be thought to destroy this equality are ( 1 ) the centrifugal force in the rotating column of air between the vent-holes and the outside of the cap ; ( 2 ) the pressure difference necessary to exhaust the products of combustion through the holes ab the proper rate ; ( 3 ) an initial delay while the interior of the cap is filling up to the true pressure outside .
The effect of ( 1 ) is easy to calculate .
In the case of ( 2 ) the maximum rate of efflux required is known from the composition of the powder and rate at which it is burnt .
Observations were made of the pressure differences required to attain a measured rate of efflux with the actual caps .
These observations also apply to ( 3 ) .
As a result it appears that the total pressure difference will only amount to more than atmospheres for a very short time ( of the order of 1/ 20 sec. ) at the start .
The effect is therefore completely negligible .
A far more serious question is , whether the pressure on the holes is actually the same as the pressure on the surface of the cap in the absence of the holes , the holes themselves being small .
This is an accepted principle in aerodynamical work at low velocities , there seems no valid reason to expect it to fail at higher velocities .
As a control of this assumption , the comparative series and were fired ; in the diameters of the holes were twice those of C. There is definite evidence that the dynamic pressures determined as above are different for these two series .
The difference is not appreciable at the higher velocities , as the times of burning at setting ( 3 ) are the same , but for values of between and the pressure determined by is distinctly less than the pressure determined by ( see fig. 2 ) .
The holes of the series are therefore not " " smali holes\ldquo ; at these velocities .
It thus remains somewhat doubtful whether the holes of series are really " " small If they are not , then the true dynamic pressures are probably somewhat higher .
The difference is likely to be small , but it is not possible to assign to it a numerical upper limit .
The best direct experimental evidence will be found in the following papers : ( 1 ) G. Fuhrmann , " " Theoretische und experimentelle Untersuchungen an Ballon1nodellen 'Jahrbuch der Motorluftschiff-Studiengesellschaft , ) vol. 6 , p. 12 ( 1911 ) ; ( 2 ) B. M. Jones and C. J. Paterson , " " Investigation of the Distribution of Pressure over the Entire Surface of an Aerofoil Advisory ComnUttee for Aeronautics .
Reports and Memoranda , No. 73 , March , 1913 .
Pressure Distribution on Shell Moving High Velocities .
217 The remaining points are less important .
We have tacitly assumed that , throughout the motion , the axis of symmetry of the shell and the direction of motion of its centre of gravity coincide .
The departures from this position of symmetry will be very small .
They may exceed 1o or initially for a short time , but for most of the motion they will be much less .
The effect of such departures on the average pressure inside the cap has been detennined at low velocities and is known to be very small indeed .
The whole effect will probably be igible here .
Finally , we come to the uncertainties in the behaviour of the fuse , and especially in the position of the zero point .
All the fuzes used were subjected , as far as possible , to precisely similar treatment .
There is no room for any suspicion of systematic differences between different parts of the material used .
Random differences alone should remain , and the extent of these is shown by the average residuals of Table II .
It will be observed also that a number of rounds are omitted from the means and mean differences of this Table .
Since all these omitted rounds are short , and since they occur in all series except with its large holes , some occasional mechanical trouble , such as the blocking of the holes by slag or paint , is suggested .
On the other hand , an error in the zero point will cause systematic errors in all the dynamic pressures at the there velocities .
This is shown by where it is obvious that the precise position of will seriously affect the slope of the time-fuze setting curve for short times .
It is by the slopes of such a curve that the rates of ; are obtained .
Captain J. W. Bispham .
Distribution of the the velocity of sound .
It must not be supposed that this represents the al aocuracy of the results , especially at velocity ratios above or , where the effect of the zero point may be serious .
It does seem , however , possible to claim that the tables and curves of this paper give values of these coefficients of the true order of magnitude , and show correctly the type of change that takes place as the velocity of the shell passes through the velocity of sound .
An Determination of the Distribution of the Partial Correlation Coefficient in of Thirty .
By Captain J. W. BISPHAM , E. Communicated Prof. AIartin , Received November PART I.\mdash ; Samples an Uncorrelated Universe .
1 .
IntroductorlJ In .
" " Stndent\ldquo ; dealt experimentally with the distribution of the total correlation coefficient of small samples .
In particular , he dealt with values of as low as 4 for the case of zero correlation in the sampled population .
In 1913 H. E. SoperT theoretically determined the mean correlation and the standard deviation of the distribution of correlations to second approximations .
In 1915 .
A. an equation for the frequency distribution of , and in 1917 , as a result of a co-operative study by H. E. Sopel , A. W. Young , B. M. Cave , A. Lee and K. Pearson , S this was reduced to stitable form for numerical manipulation , and the frequency distributions and frequency constants for samples of size ranging from to were given for values of the correlation in the sampled population from to .
The present expel.imental investigation was commenced in 1914 , but had to be put aside during the war .
It was intended to determine whether the distribution of partial correlation coefficients for samples as small as 30 showed greater dispersion than is observed for total correlation coefficients .
has shown that for normal distributions and samples the standard deviations of the distributions should be of the 'Biometrika , ' vol. 6 , p. 302 , et seq. ' Biometrika , ' vol. 9 , p. 91 , et seq. Biometrika , ' vol. 10 , p. 507 , et S 'Biometrika , ' vol. 11 , p. 328 , et seq. G. U. Yule , ' Roy .
Soc. Proc , vol. ( 1907 ) .

