]\gt ; On 's Iheory of the Impact the Pianoforte By C. , , Palit Professor of Physics , and BAVERJI , M.Sc .
, Assistant to the Palit Professor , University of Calcu tta .
( Communicated by Dr. Gill ) T. YValker , C.S.I. , Sc. D. , .
Reeeived ovember 7 , 1919 .
1 .
Intro theory of the vibrations of the string } ) ) by Kaufmann in a well-known paper* has nred prolninentl in recent discnssions on the acoustics of this ir strtlment .
It proceeds on lines different from those adopted by Helmholtz in his classical nent of the While that the elasticity of the ) ianoforte hammer is not a negligible factor , Kaufmann set out to sinnplify the lnathen ) atical analysis by its eHect iether , and the as a particle possessing only inertia without incr .
The lloCio]l .
the string following the impact of the hammer is found from the lions and from the functional solutions of the equation of wave ratio on the .
On this basis he gave a treatment of two cases : ( 1 ) particle inlpinging on a stretched string of infinite lengtl ) , and ( 2 ) a particle impinging on the centre of finite , neither of which cases is of much interest from an acoustical point of view .
The case of practical importance treated by hinu is that in which a particle impinges on the string near one end .
For this case , he gave only an theory from hich the duration contact , the motion of the point struck , and the of the vibration.curves various points of the string could be found .
There can be no doubt of the importance of vork , and it naturally becomes necessary to extend and revise his theory in directions .
In several respects , the theory vaits fuller development , especially as regards the harmonic analysis the modes of vibration set } ) by impact , and the detailed discussion of the lence of the elasticity of the hammer and of velocities of impact .
Apart from these points , the question arises whether the approximate method used by Kaufmann is sufficiently accurate for practical purposes , and whether it may be regarded as apphcable when , as in the pianoforte , the point struck is distant one-eighth * Annalen der Physik , ' vol. 54 .
' Proc. Phys. Soc London , February , 1913 ; 'Nature , ' 1913 and 1914 , and ' Proc. Royal Institution , ' 1914 ( Prof. G. H. Bryau ) .
See also D. .
Miller , ' Science of Musical Sounds , ' Macmillan \amp ; Co. , p. 207 1915 ) .
'Sensations of Tone , ' English Translation by Ellis , , and Appendix V. 100 Prof C. V. Raman and Mr. B. Banerji .
On Kaufmann 's one-ninth of the length of the string from end .
Kaufmann 's atment is practically based on the assumption that the part of the string between the end and the point struck remains straight as long as the hammer and string remain in contact .
Primd facie , it is clear that this assumption would introduce error when the part of the string lder reference is an appreciable fraction of the whole .
For the effect of the impact would obvionsly be to excite the vibrations of this portion the string , which ) continue so as the hammer is in coYltact , and would also influence the mode of vibration of tlJe string as a whole when the hammer loses contact .
mathematical theory which is not subject to this error , and which is applicable for any posit , ion of the striking point , thus seems called for .
In the present cormnunication , it will be shown how the general case of an inelastic particle impinging at any specified point on the may be dealt with ] , and the magnitudes of the forces exerted durin , impact and the duration of contact llay calculated .
To illustrate the method , a number of actual cases have been worked out nunlelically , and an attempt is made to compare the indications of theory with the results ound in experiment .
2 .
Analytical Two possible methods for dealing with the problenl under consideration suggest themselyes .
The first is the rigorous application of the functional solutions of the equation of wave propagation after the manner of aufmann taking into account the multiple reflections that occur at the particle and at the two extremities of the As differe1lt expressions have to be used for the motion on the two sides of the point , it is obvious that such a treatment would be extre1nely cumbrous , and indeed of impracticable length .
other method that ests itself is that of } ) ressing the otio that sues on impact as resultant of the inharmonic vibrations of the string having a load attached to it at the striking point so long as the }lammer is in contact with it , and thereafter as a free periodic vibl'ation of the usual kind .
Here also a difficulty arises .
For the inharmonic lnetrical series the exerted by the particle on the not uniformly conVergent in the neighbou1hood of the discontinuities in the function its sum , it is thus not practicable directly to carry out an accurate summation of its terms to enable the duration of contact between particle and string to be found .
By a judicious combination of the two methods indicated above , however , a fairly simple and straightforward process may be evolved for the of the orce exerted by the hamlner on the for any position of the striking point , and thus nining the duration of contact .
Theory of the of the noforte Hammer .
In the following analysis is the linear density of the string , represent respectively the length of the whole string and of the two parts into which the striking point divides it , and is the mass of the particle .
The free periods of the loaded string are determined by the values , etc. , satisfying the equation* .
( 1 ) The displacement at any point of the loaded string is expressed by the infinite series , ' where is the velocity of transverse waves on flJe string , and the alternative expressions refer to the two parts of the The constants , etc. , in ( 2 ) have to be found from the initial conditions .
The system is initially without velocities or displacements except in regard to the particle of mass , which has a velocity .
Using the notation adopted in of Lord 's ' Theory of Sound , ' vol. 1 , we find that the displacement of the point struck at any instant during the 'impact is given by the expression where .
( 4 ) Integrating ( 4 ) and simplifying by the aid of the relation given in ( 1 ) , we obtain The force exerted by the particle ou the is therefore iveu by the equation .
( 6 ) Equation ( 6 ) expresses the force exerted by the particle on the string as the sum of an infinite series of simple circular functions of the time , whose frequencies are the same as that of the vibrations of the loaded string , and whose relative magnitudes depend only on the ratio of the mass of the * Lord Rayleigh 's ' Theory of Sound , ' Art .
136 .
102 Prof. C. V. Raman and Mr. B. Banerji .
On 's hammer to the mass of the string , and the ratio in which the striking point ( livides the string .
As remarked above , a difficulty arises in attempting to carry out a numerical summation of the series for all values of , owing to the discontinuous nature of the function which the sum represents .
This difficulty may , however , be evaded , and the work of numerical computation reatly simplified , by finding the nitudes of the discontinuities in the function and the instants of time at which they occur directly from the principles of wave-propagation .
This may be done in the way .
At time the particle impinges on the wire , and two discontinuous changes of velocity , each equal to ( the initial velocity of the particle ) , travel out , one on each side of the , reach the ends of the in due course , are reflected , and return again , much in the same way as in the theory of the Helmholtzian vibrations of a bowed string .
When a discontinuity eaches the particle it is reflected out again without alteration of magnitude .
This follows from the fact that the particle constitutes in effect a region of infinite density on the string .
Thus , at intervals of time 2 , and again at intervals , etc. , from the commencement of the impact , we have discontinuous changes of velocity reaching and reflected from the particle .
At these instants the acceleration of the particle suffers corresponding discontinuous changes .
At the commencement of the impact the pressure exerted by the particle on the string is equal to 2 , as can be shown directly from the principle of the conservation of momentum .
Accordingly , at the instants etc. , and also at the instants , etc. , the pressure exerted suffers discontinuous increases of 2 .
( If in any case the reflected waves from the two ends arrive simultaneously at the striking point , the pressure would increase by 4 ) In the intervals between the rival of the reflected waves the pressure exerted by the particle would decrease in a continuous manner defined by equation ( 6 ) above .
Now the question is , what would the series in ( 6 ) represent at the points of discontinuity , that is , when is put equal to 2 , or 2 , etc. ?
It is clear that by taking a sufficient number of terms , we should obtain a sum which converges to the mean value of the function at these points , and for this purpose a much smaller number of terms ( say 10 or 12 ) would suffice than would be necessary if we attempted to compute the form of the function itself in the neighbourhood of the discontinuities .
Having ascertained the mean value of the function at a point of discontinuity , its two actual limiting values follow at ouce by adding and subtracting half the nitude of the discontinuity , that is , .
( If two reflections arrive simultaneously at the striking point in any case , have respectively to add and subtract 2 Theory of the Impact of the Pianoforte The graph of the continuous part of the function may be then filled in .
free hand by joining up the alternate ends of successive discontinuities by smooth curves .
3 .
merical Cof the Force b , y To illustrate the application of the foregoing theory to actual cases , the authors have carried out a numerical computation of the magnitudes and frequencies of the components of force exerted by the impinging particle on the string for a particular ratio of their masses , and for 27 different positions of the iking point situated at intervals the string between its end and the centre .
In the light of the practical experience ained , it may be useful to indicate brietiy how the work may be so as to secure sufficient accuracy with the minimum of labour .
* The first step is the deter1uination of the values of , which satisfy equation .
If the mass ' of the hammel be zero , these values are , etc. On the other hand , if be very large , tends , and the higher roots fall into two groups which tend to the limits etc. , and , etc. , respectively .
For any actual value of , the lowest root of equation ( 1 ) has to be found by trial .
The work be lightened when it has to be done for a large number of positions of the striking point , by first finding by trial for three or four positions of the striking point at wide intervals on the string , and then finding it for the other points by graphical interpolation and testing the values thus obtained by actual substitution in the equation ( 1 ) .
The calculation of the higher roots of this equation is a simpler task .
For , to a first approximation , is equal to either or , where and are integers .
To a second approximation , is equal to or , plus a correction which is either ) or .
( In special cases , where , the correction to the value of is The values of , etc. , thus obtained are generally sufficiently accurate , but the values of may require further improvement by actual test against equation ( 1 ) , especially when the ; point is close to the end of the string .
The second step is the determination of the magnitudes of the components of force from equation ( 6 ) .
The fact that and appear both in equation ( 1 ) and in ( 6 ) , helps to save some labour .
Except when one of these quantities , that is either or , is generally snlall in comparison with the other , and the magnitude of the component of force in ( 6 ) is practically determined by the smaller of the two quantities , which alone Durgadas Banerji , arrying ounlerical workThe authors wssistance t. received from 104 Prof. C. V. Raman and Mr. B. Banerji .
On 's requires to be determined with accuracy .
This may be found from ( 1 ) , so as to satisfy this equation , the approximatvalue of ( ascertained as indicated in the preceding paragraph ) used for the computation of the other quantities in the equation .
The results thus obtained may be substituted in the denominator of ( 6 ) , and enable the force-components to be accurately determined .
It is of interest to notice the manner in which the computed magnitudes thd frequencies of the components of force are found to depend on the of the striking point .
Since the frequencies of the components are those of a string carrying a load at the point of impact , they form a series of hich the terms ( in accordance with a well-known principle ) are separated by the natural frequencies of the unloaded string .
The frequency of the component of force is thus intermediate between that of the and harmonics of the string , and is a maximum when the striking point is at a node of the harmonic and a minimum when at a node of the harmonic .
The nitudes of the components of force similarly fluctuate , but to a much larger extent than their frequencies , and in the opposite direction .
Thus the component force is zero when the striking point is at a node of the harmonic , and is a maximum when it is a node of the harmonic , being largest when the node is that nearest the end of the string .
For example , the ninth component of force is zero when the striking point is at , rises rapidly to a large maximum when it is shifted to and falls quickly to a small fraction of this value when it is further shifted to or or .
It shows similar fluctuations when the striking point is moved still nearer the centre of the string , becoming zero when the striking point is at , and reaching large values when it is at or , but these maxima are not so ) reat as the first .
The manner in which the frequencies and nitudes of the components of force change when the striking point approaches very near one end of the string is specially worthy of notice .
Here the first component ( which elsewhere is generally much larger than the rest ) falls off rapidly in nitude , and as the striking point is brought nearer and nearer the end , the second , third , and higher ponents increase , and become in succession the largest in magnitude , and each in turn then falls off , giving precedence , as it were , to the next component in the series .
The maximum value of each component is reached when its frequency falls to the minimum value and begins again to rise steeply .
Theory of the )of the noforte H 4 .
of the of Contact .
Using the computed values of the and frequencies of the collponents of force , the graph the discontinuous fluctuations time of the force exerted by the impiugparticle on the string may be plotted in the manner explained in a previous section .
The point at which the cuts the axis of ( time gives at once the duration of contact .
Figs. 1 and 2 represent the force-time ooraphs for the two cases in which the striking point is at and at respectively .
These diagrams are fiven to illustrate the manner in which the form of the graphs and the duration of contact alter as the striking point is removed further and further from the end .
In both of these , the hamlner loses contact with the long before the by the mpaet has had time to reach the farther end of the string and to return after reflection to the striking point .
The discontinuities appearing in the graphs accordingly represent the siye reflections of the waves between the striking point and the njarer end of the .
It will be noticed that in fig. 1 12 discontinuities appear which are close ogether , and in fig. 2 there are only six which further apart .
The effect of rethe striking point further from the end of the is to the duration of tcontact in a continuous manner , so long as the number of discontinuities in the graph emains unaltered .
At the successive stages , however , at which the number of discontinuities in the graph decreases by unity , the duration of contact decreases by a finite amount in a discontinuous manner , these jumps being the greater and apart , the more distant bhe striking point is from the end .
It is thus clear that the duration of contact is a discontinuous function of the position of the striking point .
When the striking point is so far emoved from the end of the string that reflections from both ends of the string have to be taken into account , then 106 Prof. C. .
Raman and Mr. B. Banerji .
On 's either a decrease or an increase in the duration of contact take place in a discontinuous manner : the former when the number of discontinuities in the graph goes up by unity with the change in the position of the point , and the latter when it goes down by the same amount .
The whole course of values for the duration of contact for 27 different positions of the striking point between the end and centre of the string is shown in fig. 3 Theory of the lmpact of the Pianoforte Hammer .
The duration of contact found by this method rees exactly with that found from Kaufmann 's treatment for the case in which the striking point is at the centre of the .
There is also sensible agreement of the duration of contact with that found by the approximate method of Kaufmann when the iking point is sufficiently close to the end of the string ; say at or 1/ 20 .
Elsewhere , Kaufmann 's approximate treatment fails .
Figs. 1 and 2 also illustrate another feature which is vorthy of notice .
The dotted lines in the two rams represent the graphs of the force exerted by the particle on the string , if the more rapid periodic fluctuations which may be arded as due to the vibrations of the segment of the string are glected .
It is seen that the two dotted curves are distinctly unsymmetrical in shape and follow in fact the outline of a damped harmonic curve .
This in agreement with the approximate theory given by Kaufmaun .
5 .
Results .
The result indicated by the foregoing theory that the duration of contact is a discontinuous function of the position of the iking point appeared worthy of an experimental test .
This was attempted in the following wa A steel wire .
long , was nickeled and silver-plated so as to ensure its surface ( iving good electric contact , and stretched over the of a sonometer .
The linear density of the wire was .
per centimetre .
A small solid brass cylinder was mounted at the end of a pivoted shaft , and could be caused to transversely on the wire .
Immediately on impact , the cylinder and shaft of the hammer completed an electric circuit through the wire and a sensitiye Leeds and Nortbrup ballistic galvanometer .
Three or four readings of the throw of the galvanometer were taken for each of a number of } ) ositiollS of the striking point , and the average struck .
To secure that the variable electric resistance at the point of contact should not influence the results , a non-inductive resistance of half a megohm was included in the circuit .
The results for two different of the cylinder have been plotted in fig. 4 .
It is obvious from the experimental curves that the duration of contact does not continuously increase as the striking point is removed further and further from the end of the but that it is subject to rapid fluccuations , the nitude of which increases as the striking point is moved away the end .
Of the two curves shown in fig. 4 , the upper relates to the case of an impinging cylinder having a mass of 24 , and the second of a cylinder having a mass of 16 , the figures in each case including a correction for the inertia of the shaft calculated after the manner of Kaufmann .
108 Prof. C. V. Raman and Mr. B. Banerji .
On Kaufmann 's ratio of the mass of the hammer to ths mass of the string for the upper of the two curves in fig. 4 is the same as thAt for which the theoretical computations have been carried out and plotted in fig. 3 , and the general resemblance between the computed and observed curves is obvious .
Why there is not a much closer agreement is an open question .
It must be remembered that , in man .
respects , the experimental arrangements do not strictly reproduce the conditions assumed in the theoretical calculations .
The finite size of the cylinder and elastic flexure of the shaft , the stiffness of the wire and its yielding at the ends , and the effect of ( yravity on the motion of the impinging cylinder , are factors which probably influence the results in an appreciable degree .
There is no doubt , however , that the experimental results shown in fig. 4 , broadly speaking , confirm the correctness of the theoretical results , and the suitability of the method of calculation set out in the paper .
From acoustical point of view , a further development that would be of interest is the theoretical determination ( and comparison with experiment ) of the manner in which the amplitudes of the fundamental and higher harmonic components of the vibration of the string excited by impact depend on the mass of the hammer and the position of its striking point Theory of the Impact of the Pianoforte Hammer .
over the entire possible range .
* It may be remarked that equations and ( 5 ) of the paper determine the motion at every point of the string during the continuance of the impact .
The motion , after the hammer has left the , would have to be separately determined .
This could no doubt be found from the known displacements and velocities at various points of the string at the instant the hammer leaves it .
One of doing this is by a geometrical method similar to that used by Kaufmann .
A better method , however , would be to express the motion resulting from impact in terms of the forces exerted by the impinging particle and the duration of contact .
The graphs from which the duration of contact is deterlnined would be very useful in this connection , as the integrations could be carried out mechanically with the aid of these curves .
6 .
and Conrtusion .
The present paper is chiefly concerned with a revision and extension of the theory of the impact of the pianoforte hammer developed by Kaufmal ) It is pointed out that the approximate method used by Kaufmann , in which he assumes that the part of the string between the striking point and the nearer end remains straight so as the hammer in contact is unsatisfactory , and cannot be regarded as applicable when this part of the string forms an appreciable fraction of the whole .
It is shown how this assumption may be dispensed with , and general case , in which an inelastic particle impinges at any point on the stl.ing , may be dealt with rigorously and subjected to numerical computation .
The manner in which the force exerted by the impinging particle varies is determined , partly from the functional solutions of the equations of wave-propagation , and partly from the theory of normal vibrations .
Numerical computations have been carried out of the components of force for a particular ratio of the masses of the string , and of the impinging particle , and for 27 different positions of the striking point , and the duration of contact is deduced by a graphical method .
An interesting result obtained is that the duration of contact does not continually increase as the striking point is removed away from the end of the string , but is subject to discontinuous fluctuations .
Experimental work , broadly speaking , confirms this indication of theory .
The present paper is preliminary to a more complete investigation , in which the authors hope to develop fully various points in the theory of the The choice of the position of the striking point actually adopted in the pianoforte is no doubt determined by its influence on the useful effect produced by impact of the hammer .
See G. H. Berry , ' Phil. Mag April , 1910 .
Arts .
128 and 130 , Rayleigh 's ' Theory of Sound ' vol. 1 .
Mr. S. D. Carothers .
pianoforte , especially ( 1 ) the manner in which the amplitudes of the component partial vibrations of the string pend on the position of the striking point ; ( 2 ) the effect of the mass and elasticity of the hammer ; ( 3 ) the characteristics of the coupled vibrations of the string and soundoard of the pianoforte ; ( 4 ) the manner in which these vibrations decay by communication of energy to the atmosphere ; and ( 5 ) the theoretical tion of the quality of pianoforte tone .
Further work on these points , and experimental tests on an actual pianoforte , are now in progress .
In conclusion , the authors wish to thank Dr. Gilbert Walker , F.R.S. , who has taken a lively interest in } work , for his kindness in it before the Royal Society .
Plane : the of Stress .
By S. D. CABOTHEBS , Assoc. M.Inst .
C.E. ( Communicated by Prof. A. E. H. Love , F.R.S. Reoeived December 9 , 1919 .
) ( 1 ) advantages of the direct determination of the stresses in an elastic solid have been pointed out Prof. J. H. lIichell* .
The principal line of attack in the case of plane strain has been by aid of the well-known stress function method , by which the stresses are determined from a single stress function , of and , satisfying where ( 1 ) The lines along which further advance might have been expected , and the difficulties which have been met with , are discussed by Prof. A. E. H. Love .
It seems to the writer , however , that a point of very considerable importance has been overlooked , viz. , that the stress function method gives a set of stresses which can in most cases be resolved into two distinct sets , each of which leads to strains satisfying the identical relations between the strain components .
For example , if is a plane harmonic function , satisfies , and the stresses derived from this by the well-known formulae , and , ( 2 ) 'Proc .
Lond. Math. Soc vol. 31 , p. 100 .
Math. Theory of Elasticity , ' 2nd Edition , p. 211 .

