520 Mr. A. J. Ellis .
[ June 17 , June 17 , 1880 .
THE PRESIDENT in the Chair .
The Presents received were laid on the table , and thanks ordered for them .
Prof. Charles Niven and Dr. William A. Tilden were admitted into the Society .
The Right Hon. Alexander James Beresford Hope , whose certificate had been suspended as required by the Statutes , was balloted for and elected a Fellow of the Society .
The following Papers were read :\#151 ; I. " Notes of Observations on Musical Beats .
" By Alexander J. Ellis , B.A. , F.R.S. , F.S.A. Received May 28 , 1880 .
During the last three years I have been greatly occupied with observing and counting musical beats , for the purpose of discovering the cause and amount of error in Appunn 's reed tonometer , and of measuring the number of vibrations made in a second by tuning-forks and organs , as materials for my " History of Musical Pitch .
" * The following are brief notes of some of the observations then made:\#151 ; When two musical notes nearly but not accurately form a consonance , or are in unison , they beat .
Under ordinary circumstances the number of beats in a second of a disturbed unison is equal to the difference of the number of double vibrations in a second made by each note .
It is not so always , as will be shown later on .
If x and be the " pitch " or number of vibrations in a second , made by two musical tones , of which y is the sharper ; then , if my\#151 ; 0 , the tones form what I have termed a cothat is , the \#171 ; th partial of x falls on the same rank or seat as the m th partial of y. Considences are not always consonances , because other partials of the notes may beat roughly , as when m : n= 8 : 9 or 9:10 or 15 :16 , which are well known dissonances , but give appreciable considences .
But if the pitch of either xor ybe slightly altered , so that my\#151 ; the two consident partials become what I have termed , or placed on different ranks or seats , and b beats in a second are heard , being called " sharp " when positive , that is when my\gt ; nx , and " flat " when negative , that is when my\lt ; nx .
This includes the unison for which * A paper read before the Society of Arts on March 3 , 1880 , and printed in their Journal for March 5 , with an appendix on April 2 , 1880 .
1880 ] Notes of Observations on Musical Beats .
m\#151 ; n. Hence all beats heard are beats of simple partial however compound may be the tones which contain them .
This agrees thoroughly with my observations .
Tuning-forks are comparatively simple but always possess an audible second partial or octave , and sometimes higher partials still , capable of being so reinforced by resonance jars properly tuned to them , that beats can be separately obtained from them and counted .
This is a matter of great importance in the construction of a tuning-fork tonometer .
When the tone is very compound , as in the case of bass reeds ( especially those of Appunn 's tonometer , furnished with a bellows giving , when properly managed , a perfectly steady blast for an indefinite length of time ) , beats can be obtained and counted from the 20th to the 30th and even the 40th partial , without any reinforcement by a resonance jar .
Taking tuning-forks first , I find it advantageous to hold the beating forks over one or two resonange jars , tuned , by pouring in water , to the pitch of the partial to be observed , whether it be the prime of both or the prime of one and the second ( or octave ) of the othei .
There may be small differences , but I have not found any difference appreciable by my methods of observation in the number of beats in a second , whether the resonance jar is the same or different for the two forks , and whether it is exactly or very indifferently tuned to each fork , but a tolerably accurate tuning much improves the tone and length of the beat .
In that case the resonance jar practically quenches all other partial tones , and the beats are distinctly heard as loudnesses separated by silences .
If no jar is used , the other partials are heard .
In the case of the octave , the low prime becomes a drone and fills up the silences .
In the case of beating primes , the octaves , which are beating twice as fast , tend to confuse the ear .
Sometimes the second partial of a fork is so much stronger than the prime , that when the fork is applied to a sounding-board , only the octave is beard , which is inconvenient .
This is entirely avoided by the resonance jar .
Beats being a case of interference , the amplitude of the beating partials should be equalised as much as possible .
With two forks of very different size and power , it is easy to regulate the amplitude by holding the louder fork further from the jar .
Otherwise the beats become blurred and indistinct .
For powerful reeds or organ pipes , beating with forks , it is best to go to a considerable distance from the reed or pipe and hold the fork close to the ear or over a jar .
I find o0 or 40 feet necessary for organs ; in Durham Cathedral , where the pressure of wind was strong and my forks weak , I found 60 or / 0 feet distance much better .
As I was not able latterly to go to a distance from Appunn 's reed tonometer , having to pump it myself , I found it impossible to count the primes of the upper reeds by the octaves of my forks , which were completely drowned .
Mr. A. J. Ellis .
[ June 17 , I find beats of all kinds most easy to count ( by the seconds hand of a chronometer ) when about 4 in a second .
They can be counted well from 2 to 5 in a second .
Above 5 they are too rapid for accuracy ; below 2 and certainly below 1 , they are too slow , so that it is extremely difficult to tell from what part of the swell of sound the beat should be reckoned .
Partly from this reason , perhaps , I have found great variety in counting successive sets of such slow beats .
I never use beats of less than one in a second , if I can avoid it .
When the beats are slow it is difficult to discover by ear which of the two beating tones is the sharper ; and even fine ears are often deceived .
It is easy to discover , however , by putting one of the forks under the arm for a minute .
This heats and flattens it by 2 or 3 beats in 10 seconds .
Hence if the beats with the heated fork are slower , it was sharper , because it has been brought nearer the other ; if faster , it was flatter and has been brought further away .
Count for 10 , 20 , or 40 seconds , according to the fork .
Up to 20 or 30 beats in 10 seconds it is easy to count in ones , but from 30 to 50 it is best to count in twos , as oue-ee , two-ee , \amp ; c. , beginning with and hence throwing off one at the end .
When counting for 20 seconds I always count in twos , and for 40 seconds in fours , as one-ee-ah-tee , two-ee-ah-tee , \amp ; c. , because I have to divide the result by 20 or 40 ; and this division is avoided by the count itself .
Owing to difficulties in beginning and ending the count , I find the possible error per second to be 2 divided by the number of seconds through which the count extends ; and that it is best to take a mean of 5 to 10 counts for each set of beats .
Temperature plays an important part .
Forks should not be touched with the unprotected hand ; they otherwise easily flatten by 2 beats in 10 seconds .
Interpose folds of paper .
I use two folds of brown paper stitched between two pieces of wash-leather .
Large forks are generally on resonance boxes and need not be touched , otherwise the same precautions should be used , as they are very sensitive , and retain the heat longer than small forks .
Scheibler 's forks are fitted with wooden handles .
In tuning , the file heats and flattens ; the result , therefore , can seldom be known for a day or two , when the forks have cooled and " settled , " as they will be sure to " jump up .
" I find it best to leave off filing when the forks are two or three tenths of a vibration too flat .
In sharpening there is , therefore , great danger of doing too much , as the fork remains apparently at the same pitch , the flattening by heat balancing the sharpening by filing .
Hence all copies should be compared some days after , by means of a third fork about four vibrations flatter or sharper than each , to avoid the slow beats of approximate unisons .
The filing also seems to interfere with the molecular arrangement of the forks .
The thermometer should be always consulted when beats are taken .
But if the beats are between two forks , of which the pitch of one at a 2880 .
] Notes of Observations on Musical Beats .
523 given temperature is known , and both forks may be assumed to be altered in the same ratio by heat , then the temperature need not be observed ; but the unknown fork may be presumed to be as many vibrations sharper ( or flatter ) than the measured fork at the temperature at which the latter was measured , as beats in a second were observed to take place .
This is because the alteration is very small , and would be quite inappreciable for the few vibrations between them .
But for tonometrical purposes an allowance must be made .
The coefficient of temperature has not been satisfactorily determined .
It varies from '00004 to '00006 for each vibration and each degree Fahrenheit .
Possibly the mean '00005 is the best number to take , but I have used 1-^21000='0000476 .
The flattening seems to be chiefly due to the effect of temperature on elasticity .
A large fork of about 435'44 vibrations at 59 ' F. , grew sharper and beat more and more brightly as the temperature descended to 10 ' or 15 ' F. , being easily counted for 20 seconds and more .
At 104 ' F. it could scarcely be counted for 20 seconds , at 112 ' F. scarcely for 10 seconds , and at 164 ' F. I could not count it at all , the sound not lasting more than 2 or 3 seconds , and the beats varying during that time .
The effect also seems to vary with the metal and make of the fork and its size , and the coefficient to be greater for high than for low temperatures .
For a fork of about 256 vibrations at 59 ' F. , the coefficient from 14 ' to 59 ' F. was about *0000305 or l-f-32,760 , but from 59 F. to 175 F. about '0000548 or 1-*-18,280 .
These experiments , which I made by dipping the forks in freezing mixtures and hot water , and beating them with a fork at mean temperature , are unsatisfactory .
In the same way M. Aristide Cavaille-Coll , the well-known organ-builder of Paris ( private letter ) , experimenting on two forks of about 435 vibrations , one of Scheibler 's and one a large-sized diapason normal , found Coeff .
for 1 ' F. and 1 rib .
1st fork , from 59 ' F. to 194 ' F. . .
*0000567=1\#151 ; 17,650 2nd " " 60''8F .
to 194 ' F. . .
'0000589=1-^16,970 Scheibler ( " Tonmesser , " p. 50 ) himself found for a rise of 45 F. Coeff .
for 1 ' F. and 1 rib .
1st fork , about 440 vibrations . .
*00006=1-5-16,670 2nd " " 220 " . .
*00005=1-5- 20,000 Kayser , for two large forks , furnished with mirrors , \amp ; c. ( " Wied .
Ann. , " 1879 , p. 444)\#151 ; Coeff .
for 1 ' F. and 1 vib .
1st fork , about 72 vibrations . .
'0000494=1 -4-20,250 2nd " " 85 " . .
*0000566=1-5-18,000 Koenig 's recent careful experiments ( " Wied .
Ann. , " 1880 , p. 413 ) , not made by freezing mixtures and hot water , give two different coMr .
A. J. Ellis .
[ June IT , efficients , deduced originally from a fork of 64 vibrations and extended to one of 256 .
For 1 ' C. and 1 vib .
For 1 ' F. and 1 rib .
1st or general coefficient . . .
1-4-8943 ... .
'00006212 = 1-7-16,097 2nd or particular " ... 1-4-8951 ... .
'00006207=1-4-16,112 These are practically the same , but it is the latter which he uses in his reductions .
Professor McLeod by his machine ( " Proceedings , " 1879 vol. xxviii , p. 291 , and " Phil. Trans. , " 1880 , p. 1 ) at first obtained a result nearly identical with Koenig 's , but afterwards , when experimenting on one of my forks of about 440 vibrations , he found ( private letter ) the coefficient to be '00004882=1-4-20,490 .
While Professor Alfred Mayer ( private letter ) by repeated experiments on Koenig 's and other forks , exposed to the cold of American nights , and counted by beats with forks at mean temperature , obtained the coefficient '00004545= 1-4^22,000 , and I adopted 1-4-21,000 as a mean of those of Professor McLeod and Professor Mayer .
The error must certainly be very small for all Scheibler 's forks .
For organ pipes where great exactness does not seem to be possible , I find the coefficient '00104 for 1 ' F. and 1 vibration quite sufficiently accurate and to give concordant results generally .
In some organs where the air for blowing is cooler than the air of the room , this is too large , according to the observations of Mr. A. J. Hipkins , at St. James ' Hall , the coefficient is nearer '0005 ( private letter ) .
In taking the pitch of organ pipes by measured forks , I usually neglect the small alterations of the pitch of the fork , and allow fully for the change of pitch in the pipe .
It is probable that this coefficient does not answer for the stopped and fancy pipes .
I have applied it only to open , and generally metal cylindrical pipes .
To construct the fork tonometer , invented by J. Heinrich Scheibler ( b. 1777 , d. 1837 ) , a silk manufacturer at Crefeld , obtain a set of about 70 good forks ' with parallel prongs , and of a tolerably large size ; tune the lowest to about the C ( or B for English high pitch ) between the bass and treble staves of any organ or piano , and tune the rest roughly each about four beats in a second sharper than the preceding .
Then fit them with wooden collars or handles , and allow them to rest for three months , if possible in the same temperature at which they will be counted .
This was not the process adopted by Scheibler , but is much simpler .
Then count the beats between each set most carefully , at a temperature which remains as uniform as possible .
It may be necessary to use a high temperature ; thus Scheibler 's was from 15 ' R. to 18 ' R. = 65'75 to 72''5 F. , which I reckon at 69b F. as a mean ; and Koenig now works at 20 ' C.=68 ' F. , but announces that his former 256 vibrations was only correct at 26''2 0 .
= 79''16 F. ( " Wied .
Ann. , " 1880 , p. 413 .
) Count on one day the beats between forks 1 and 2 , 3 and 4 , \amp ; c. , and on 1880 .
] Notes of Observations on Musical Beats .
525 another between forks 2 and 3 , 4 and 5 , \amp ; c. , so that the same fork is not used for two counts on the same day .
Excite by striking with a soft ball of fine flannel wound round the end of a piece of whalebone , as a bow is not convenient unless the forks are tightly fixed .
Each blow or bowing heats , and hence flattens , and this tells if the experiments on any one fork are long continued .
Count each set of beats for 40 seconds if possible , and many times over , registering the temperature and the beats , and take the mean .
Scheibler counted by a graduated metronome , set constantly to an astronomical clock , when the weight was at 60 , to eliminate the effect of temperature , and he altered the position of the weight ( finally by a micrometer screw ) so that there were always four beats to each swing of the pendulum .
He seems to have attained extraordinary accuracy .
Having counted all , observe those forks which are near the octave of the lowest fork .
Find two such , beating with the octave ( that is , the second partial tone ) of the lowest fork less than they beat with each other .
Then the sum of all the beats from the lowest fork to the lower of the two forks , added to the beats of the octave ( that is , the second partial tone ) with that fork , is the pitch of the lowest fork .
Hence the pitch of all the forks is known .
The extra high forks are for verifying by the octaves of several low forks , and for the purpose of subsequently measuring .
From such a tonometer any other can be made , and the value of each fork at another temperature calculated .
Scheibler made a 52-fork tonometer with infinite trouble , on another plan , and counted it with marvellous accuracy .
This tonometer , which I have made many efforts to find , has absolutely disappeared and his family knows nothing of it .
But he left behind him a 56-fork tonometer , believed to proceed from 220 to 440 vibrations , and through the kindness of Herr Amels , an old friend of the Scheibler family , who obtained it from Scheibler 's grandson , I have had the use of it for a year .
I had to count it as well as I could , just as if it had been a set of forks such as I have described , and I found it was not what was thought , but that only 32 sets of beats were 4 m a second , and the other 23 sets varied from 38 to 42 in 10 seconds .
I found also that the extremes were probably of the same pitch as in the original 52-fork tonometer .
After then counting it as well as I could , and obtaining 219*27 vibrations in place of 219*67 , at 69 ' F. , I distributed the error of 4 beats in 10 seconds , as 2 in 100 seconds , among 20 of the 23 sets which were not exactly 4 beats in 10 seconds , leaving the first 3 sets , which I had repeatedly counted and felt sure of , unaltered .
Then I reduced all the values from 69 ' to 59 F. Finally to verify my res I measured by beats with Scheibler 's forks as thus determined ; first 5 large forks of various pitches , which I had had made for me in Paris , and then 4 forks of Koenig 's belonging to Professor McLeod .
Professor McLeod himself kindly measured all of them , also , by his Mr. A. J. Ellis .
[ June 17 , machine , and Professor Mayer , of the Stevens Institute , Hoboken , New Jersey , U.S. , kindly measured the first 5 forks by his electrographic method , both with the greatest care and precaution .
The three sets of measurements agreed to less than 1 beat in 10 seconds , and more often less than 1 beat in 20 seconds , when reduced to the same temperature .
Thus the value of the tonometrical measurement by beats only , and the possibility of counting a tonometer sufficiently , was fully established .
Koenig 's measurements of his own forks reduced to 59 ' F. and of the actual Diapason Normal at the Conservatoire , Paris , intended to be used at the same temperature , also agree with mine within less than the same limits .
By these forks I have counted 75 standard forks of Messrs. Yalantine and Carr , music smiths , 76 , Milton Street , Sheffield , who are thus in a position to make small copies , probably not more than half a vibration wrong at most , at a cheap rate .
When forks are counted without a resonance jar , they should not be applied to a sounding board , or held one to one ear and one to the other , but should both be held about six inches from the same ear , and their strengths should be equalised by holding the weaker fork closer to the ear than the stronger .
When the forks are screwed on and off a sounding board or resonance box , there is great danger of wrenching the prongs , unless they are held below the bend , but I have constantly seen this precaution neglected .
A wrench immediately affects the pitch and duration of sound of a fork , and renders it comparatively worthless .
Such cases have come within my observation .
The next enemy to be guarded against is rust .
Forks should be kept dry , and occasionally oiled with gun-lock oil .
Bust towards the tip affects the fork much less than rust at the bend .
My observations and experiments show that errors from rust can scarcely exceed a flattening of 1 vibration in 250 , and are generally very much less .
But as the amount is uncertain , rust spoils a fork for accurate tonometrical purposes .
My observations on reed tones are confined to those in Appunn 's tonometers at the South Kensington Museum , the Museum of King 's College , London , and Lord Rayleigh 's , where there are copies , all of which I have counted .
They consist of oblong boxes containing the reeds placed side by side .
The wind pumped into a large reservoir , is driven from it by a spring , and received into this box , the heavy lid of which , separated from the body by the usual bellows-folds of leather , presses on the wind and drives it on the reeds with a very constant pressure .
Below each reed is a pallet which , when pulled out by a valve to its full extent , allows the reed to sound at its highest pitch to which it is tuned .
If the pallet be slid in somewhat , the pitch can be flattened by as much as two vibrations without much affecting the quality of tone , and even as much as three vibrations with considerable 3880.1 Notes of Observations on Musical .
527 loss in quality .
This power of reducing the pitch slightly and instantly restoring it , is of great service in experiments upon confidence and dissidence , which the relations of the pitch of the reeds allow of being tried in a very large number of cases .
It also enables considences to be rendered perfect when the instrument , as is necessary generally the case , is slightly out of tune , as the intervals can be made closer by flattening the upper , and wider by flattening the lower reed .
The nominal values of the reeds are .
as follows : Bass tonometer , 57 reeds , numbered by their nominal value m double vibrations 8 , 9 , 10 , 11 , 12 , 18 , 14,15\#151 ; 16 , 17,18 , 19 , 20 , 21 , 22 23 24 25 26 27 , 28 , 29 , 30 , 31\#151 ; 32 , 34 , 36 , 38 , 40 , 42,44,46 , 48 , so ' 52 , st 56 , 58 , 60 , 62-64 , 63 , 72 , 76 , 80 , 84 , 88 , 92 , 96,100,104 , 108 , 112 , 116,120 , 124 , 128 .
_ Tenor tonometer , 33 reeds , numbered 0 to 32 , nominal value\#151 ; 4 X number +128=128 , 132 , 136 , 140 , 144 , 148 , 152 , 156 , 160 , 164,168 , 172 , 176 , 180 , 184 , 188 , 192 , 196 , 200 , 204 , 208 , 212 , 216 , 220 , 224 , 228 , 232* 236 , 240 , 244 , 248 , 252 , 256 .
Treble tonometer , 65 reeds , numbered0 to 64 , nominal value =4 X number +256=256 , 260 , 264 , 268 , 272 , 276 , 280 , 284 , 288 , 292 , 296 , 300 , 304 , 308 , 312 , 316 , 320 , 324 , 328 , 332 , 336 , 340 , 344 , 348 , 352 , 356 , 360 , 364 , 368 , 372 , 376 , 380 , 384 , 388 , 392 , 396 400 , 404 , 408 , 412 , 416 , 420 , 424 , 428 , 432 , 436 , 440 , 444 , 448 , 452 , 456 , 460 , 464 , 468 , 472 , 476 , 480 , 484 , 488 , 492 , 496 , 500 , 504 , 508 , 512 .
The correctness of these numbers had to be proved in the first place by counting the beats .
The sum of the various sets of beats m the treble tonometer should be 256 , in the tenor tonometer 128 , m the bass tonometer , from 8 to 16 , 8 beats ; from 16 to 32 , 16 beats ; from 32 to 64 , 32 beats ; from 64 to 128 , 64 beats .
I counted the beats m the treble tonometer at South Kensington several times .
On 27th October , 1876 , each set of beats being counted for 20 seconds , I obtained 256 exactly .
From 5th to 24th September , 1877 , I counted each set of beats for a minute , and many times over , and , owing to alterations in the pitch of the reeds , the beats varied from 3*85 to 4 27 in a second , their sums being 256*27 .
On 25th and 28th September , 1877,1 again counted each set for one minute , and obtained 256*28 as the sum .
On the 10th and 12th September , 1877 , I counted the beats on Lord Rayleigh 's copy , and found the sum 256*38 .
My count of the King s College copy , 13th November to 20th December , 1877 , at two minutes for each set , gave 254*75 , and was certainly erroneous .
The beats + 6 , of a disturbed considence , as y which is supposed to be near n : m , joined with the sum of the beats d , between and a , when 6 , d , m , n are known , give y and x by the equations my\#151 ; nx = + 5 , y\#151 ; and hence the value of l , the lowest note , is known , when y l is Mr. A. J. Ellis .
[ June 17 , counted .
Of thirty determinations of l thus made for the South Kensington instrument , the highest was 256'65 , the lowest , 253-05 , and the mean , 255'85 .
Hence the value of l could apparently be not far from 256 , its nominal value .
But when compared with an TJt3 of Koenig 's , the reed beat 2'4 flat .
Supposing Koenig 's fork to have been 256 truly , this made Appunn 's 253-6 , far less than any of my counts and most of my calculations .
Lord Rayleigh and Mr. R. H. M. Bosanquet had already suggested to me that the confined air in the box of the tonometer , and the vibration of the whole instrument during the beats , " drew " the notes .
I believe they thought that it altered the pitch of the notes , and hence the number of the beats .
I began to entertain the same opinion , and devised the following experiment , the Lords of the Committee of Council on Education allowing me to remove the treble and teitor tonometers from South Kensington to the Museum of King 's College , where Professor W. G. Adams , F.R.S. , allowed me to compare these instruments with those in the Museum for many weeks .
Suppose L , M , 1ST , are adjacent reeds on one instrument , and L ' , M ' , N ' reeds of nominally the same pitch on the other . .
Practically , they were not quite of the same pitch , a circumstance which showed inaccuracy of construction .
Then I took the beats between L and M , M and N , and thus by addition obtained the internal beats between L and N , that is , those which occurred within the box of the tonometer .
When I took the beats between L and M ' , M ' and N , and thus obtained the external beats between L and N , that is , beats which were formed in the uncompressed atmosphere of the museum external to the tonometers , the number of the internal beats always exceeded that of the external .
On taking a mean of my observations , which extended to every set of beats , I found that I could reduce the internal to the external by diminishing the number of the internal beats by 76 in 10,000 .
I was not completely satisfied with the accuracy of my observations , or with what I considered the rather hazardous mean , but was unable to repeat the very long course of observations .
In November , 1879 , however , I was able to examine every reed in all the tonometers by means of Scheibler 's forks , and I found that the nominal values of the reeds in the treble tonometer could be reduced to those of the forks by a mean subtraction of 76 in 10,000 ; in the tenor tonometer ( which was altogether flatter ) , by a mean subtraction of 83 in 10,000 ; in the bass tonometer , octave 64 to 128 , by a mean subtraction of 76 in 10,000 ; octave 32 to 64 , of 64 in 10,000 ; octave 16 to 32 , of 57 in 10,000 ; the octave 8 to 16 was too uncertain to deduce a mean .
These results singularly well confirm the former .
The nominal values had been deduced from observed internal beats of reed with reed ; the external beats agreed whether they were taken between reed and reed , or reed and fork .
It is certain , therefore , that the internal beats were accelerated , whether the pitch of the 1880 " | Notes of Observations on Musical Beats .
reeds themselves was raised or not ; I think it was not .
But the nominal value , depending on the accelerated heats was , of course , increased , and the experiment of the forks shows that it was increased exactly in the ratio of the internal to the external beats .
At any rate , \ therefore , if the pitch of the tones was increased , it was not caused by t the confined air acting upon a single speaking reed , but was occasioned by the joint action of two speaking reeds .
This joint action was very powerful .
It shook the whole instrument violently .
When reeds 256 and 260 ( I cite them always by their nominal value ) were sounded together , at first only a crash was audible , but after about a second the ear could distinguish the beating of the two primes , closely resembling the beating of two tuning- forks held over resonance jars , but accompanied throughout by a great crash , which made the simple beats difficult to keep well in the ear .
This was totally different from the effect when a fork was substituted for one of the reeds .
The simple beats remained , but the crash disappeared .
Moving of the head caused considerable difference m the loudness of the simple beats , due perhaps , to placing the ear at or near a node of the sound wave , and removing it again .
It was useless to attempt to count the beats till these simple beats were well recognised .
As I ascended the scale , the crash became less .
It was much less for reeds 272 and 276 , and the contrast of these beats with those of 256 and 260 on the one hand , and 288 and 292 on the other , was very striking .
Bor reeds 320 and 324 the crash was comparatively faint .
I couldnot distinguish the beats of the second partials or octaves from the general crash arising from the beats of all the other partials .
The beats of the primes were quite separate , slow ( 4 in a second ) and distinct , and they seemed to give the time to the other beats .
For reeds 376 and 380 , the beat of the primes overpowered the crash , which became comparatively light , and after this point , the beats of the primes were always easy to find , brom reeds 412 and 416 , the beats of the primes were the principal phenomena , and after 492 and L 496 , the beats were practically simple .
In the tenor and bass tonometers , the beats were still more distressing to the ear , for even for single notes the upper partials beat clearly and slowly enough , after reed 64 , to be distinctly perceived as beats ; and the beats of two reeds sounded almost like a continually reiterated feu de joie .
Below reed 16 , the primes could not be heard at all , but down to 8 the beating upper partials could be heard .
The effect of the external beats on the ear was distinctly different from the effects of the internal beat 's .
The surge of the other partials was not so strong , and the beats were much easier to count .
At one time I placed the two tonometers fully 50 feet apart , and stood half way between them to count the beats , which were remarkably clear , the surge becoming indistinct , very like that of the distant waves on Mr. A. J. Ellis .
[ June 17 , a sea-shore , and forming a running accompaniment , totally dissimilar from the bell-like beat of the partials .
The pitch of the reeds on the treble tonometer furnishes not only numerous cases of considences , but numerous cases of disturbed consi* dences , beating four times in a second , when the reeds are in order .
Thus 3 X 256=2 X 384 , and hence 256 and 384 are a perfect considence .
But 3 x 260 2 x 388= + 4 , so that 260 and 388 form a disturbed Fifth , beating 4 times in a second , the upper note being too flat ; while 3 X 260_2 x 392= \#151 ; 4 , another disturbed Fifth , also beating 4 times in a second , the upper note being too sharp .
The fact of the upper note being too flat or too sharp is shown immediately by flattening it , as previously described ; the first beats are then made more rapid and the second more slow , but it is impossible to destroy them entirely , as the upper note cannot be sufficiently flattened .
* It is delightful , however , to take what should be a perfect considence , as all were when I first examined the instrument , and throw it out of tune by flattening either the upper or lower note , or both unequally , producing the dissident beats , and then to sharpen these notes gradually , and listen to the beats growing slower and slower till they finally entirely disappear , and then reappear as the sharpening is carried too far .
The nature of considence and dissi-dence is thus distinctly felt , and the delimitation of a considence is determined by the possibility of hearing these dissident beats when one of the extreme notes is flattened .
The beats are clear , distinct , and simple , and can be made very slow ; their pitch is also exactly what has to be expected by the number of the partials .
The other partials of the two notes in the meantime beat roughly , strongly , and very much faster than the dissident , beats .
Thus for the Fourth 4 : 3 , we may take reeds 264 and 352 , then the partials will be\#151 ; ( 1 ) ( 2 ) ( 3 ) ( 4 ) ( 5 ) 264 528 792 1,056 1,320 , \amp ; c. 352 704 1,056 1,408 , \amp ; c. ( 1 ) ( 2 ) ( 3 ) ( 4 ) Even when the 1,056 is consident , the 88 beats of 704 and 792 , and of 1,320 and 1,408 are easily heard , producing the well-known " roughness of the Fourth , while the 264 , 352 , and 528 boom along loudly and independently .
But if the upper note is flattened , the rate of the two first beats is altered , one becoming faster and the other slower , while slow beats of an entirely different character are introduced at the high pitch of about 1,056 , by the tearing apart of these formerly coincident partials .
I have watched the phenomenon over and over again for different considences , and cannot imagine a better demonstration of Helmholtz 's theories .
The following , among numerous other considences , have all been clearly delimitated by me in the way mentioned , and most of the I860 .
] Notes of Observations on Musical Beats .
forms of dissidences which , should beat 4 in a second have also been investigated : Fifth 3 : 2 , Fourth 4 : 3 , Major Third 5 : 4 , Minor Third 6 : 5 , Major Sixth 5 : 3 , Sub-Fifth 7 : 5 , Super-Fourth 10 : 7 , Supermajor Third 9 : 7 , Sub-minor Sixth 14 : 9 , Sub-minor Third 7 : 6 , Super-major Sixth 12 : 7 , Sub-minor or Harmonic Seventh 7 : 4 , Supermajor Second 8 : 7 , Major Tone 9 : 8 , Minor Tone 10 ; 9 , Small Major Seventh 9 : 5 , and Diatomic Semitone 16:15 .
The latter was most difficult , on account of the great roughness of the intervening beats , and succeeded best in the highest case , reeds 480 and 512 .
The Major Sevenths 16 :9 and 15 : 8 , however , baffled me , from the excessive roughness of the other beats .
I have frequently shown these effects to others , and as the instruments are accessible at the South Kensington Museum , they can easily be repeated.* In counting the beats of the reeds with forks , I was unable to use the octaves of the forks , as they were entirely drowned by the primes of the reeds , and hence above reed 440 I was obliged to use other forks which had been previously counted with the octaves of Scheibler 's forks ; but below reed 220 I always counted by the partials of the reeds .
By this means I was able to determine the pitch as far as 12 vib .
in a sec. , with tolerable certainty .
Occasionally I determined the pitch of a single reed by means of several partials , beating , of course , with the primes of different forks .
The following table gives the nominal numbers of some of these low reeds , with the partials used , the mean pitch determined , and the decimals of a vibration determined from the different partials , showing the close agreement of the several determinations .
Nominal number of vibrations .
Mean number of vibrations in the prime by Scheibler 's forks .
Partials used , of which the pitch was determined . .
Decimals of the numbers of vibrations of the prime as calculated from the pitch of the partials .
88 8734 3 , 4 , 5 .34 , -33 , 36 72 7146 4 , 5,6 .45 , -45 , -47 64 63-52 4 , 5,6 .52 , *52 , -53 48 47 68 5 , 6 , 7 , 8 , 9 .67 , -68 , -67 , -69 , '71 40 39-92 6 , 7 , 8 , 9 , 10 , 11 .91 , -93 , -90 , *92 , -93 , -93 36 3574 7 , 8 , 9 , 10 , 11 , 12 .73 , -73 , -73 , -735 , '745 , *74 32 31-47 7 , 8 , 8 , 9 , 10 , 11 , 12 , 13 .47 , '48 , -46 , -47 , -49 , '45 , .47 , -45 23 22-88 13 , 19 .88 , -88 16 15-94 25 , 27 , 25 .94 , -94 , -92 13 12-90 20 , 25 .89 , 91 12 11-90 20 , 28 .88 , -91 * Since this paper was sent in to the Royal Society , I have handed to the Secretary of the Science and Art Department for the use of the South Kensington Museum , a detailed account of the method of making these experiments , with tables showing how to bring the proper reeds into action.\#151 ; June 15 , 1880 .
VOL. XXX .
2 P 532 Notes of Observations on Musical Beats .
[ June 17 , After reed 12 the results were very uncertain .
After reed 32 the primes were scarcely audible , and after reed 15 they were utterly inaudible .
All that could be distinguished was the thumping of the beats of the upper partials , and these became gradually fainter and fainter , but were always audible even for reed 8 .
The mode of finding the proper forks for any partials of reeds was , simple , since the approximate pitch of the reed and the actual pitch of the fork were known .
Thus for reed 12 , the 20th partial would be nearly 20 X 12=240 , and hence would lie between the forks 239*66 and 235*69 .
On trial I found the beats to be respectively 2*00 and 1*96 ( the last a mean of several counts ) .
Then 239*66\#151 ; 20X reed 12= 2*00 , and 20 X reed 12\#151 ; 235*69=1*96 .
These give 20 x reed 12= 237*66 and 237*65 respectively , and consequently reed 12=11*88 vibrations .
The ease and certainty with which the partials could thus be picked out was delightful to observe .
As no resonance boxes or jars were used for the reeds , the objection sometimes made , that such partials are created by multiple resonances within the resonance cavity itself , falls to the ground , and the practical objective existence of the partials is established .
The practical coincidence of the values of the prime from several distkict partials shows that there was no error in assigning the pitch to the proper partial .
In the case of the four last reeds , 11 , 10 , 9 , 8 only , where successive partials are so very close , did I feel any uncertainty , and hence I have not cited these results .
It was for these cases extremely difficult to hear any beats at all , as distinct from the beats of the partials of the single notes themselves , as the partials that had to be used were very high and very weak .
From and after reed 32 there were no musical sounds at all ; indeed , even reed 64 scarcely deserved the name of a musical sound , so strong were the beats of the upper partials .
As the coefficient of temperature for reeds is unknown , a suspicion of error to a small amount attaches to all these determinations of pitch , which were made at artificial temperatures varying from 45 ' to 55 ' F. This want of correction for temperature , and liability to lose pitch from unknown circumstances , militate against the use of the reed tonometer for scientific purposes , but on account of its mumerous partials it is admirably adapted for many purposes which the stabler tuning-fork , with its small number of available partials , cannot subserve .
How stable tuning-forks are , it is difficult to say .
The lowest and highest forks of Scheibler 's tonometer do not seem to have varied , by so much as the twentieth of a vibration since 1837 , judging by my own measurements and by Professor McLeod s measurements of a fork in absolute union with the highest .
A good fork , marked 438 simple vibrations ( that is , 219 double vibrations ) m Scheibler 's own handwriting , probably about fifty years ago , is now considerably rusty , but I measure it as 218*77 double vibrations , 1880 .
] Lowering of the Freezing-Point of Water by Pressure .
533 hence it cannot have altered more than 0'23 vibration , for which the rust fully accounts .
At the same time five other forks , of a large size and very different make , after having journeyed to America and back to be measured by Professor Mayer , have , according to Professor McLeod 's measures , lost -0015 , *165 , -0205 , '0285 , and *014 vibration respectively .
The second fork was not so good as the rest , and may possibly have been slightly wrenched , as it had to be screwed in and out of a wooden holder .
The other losses scarcely exceed errors of observation and differences of estimation of the effects of temperature .
of the corroboration of Helmholtz 's theory of the objective existence of partial tones , by means of beats of these partial tones , either will i one another or with those of other compound tones .
The points to which I wish to draw attention are , the establishment the acceleration of beats which take place in confined spaces , and II .
" On the Lowering of the Freezing-Point of Water by Pressure .
" By James Dewar , M.A. , F.R.S. , Jacksonian Professor of Natural Experimental Philosophy in the University of Cambridge .
Received June 10 , 1880 .
The Cailletet pump may be conveniently employed to observe the thermal effects of compression on solid and fluid substances .
Before engaging in an investigation on this subject , it was necessary to test the apparatus , and especially the manometer .
For this purpose it seemed , on theoretical grounds , that observations on the lowering of the freezing-point of water by pressure would be a severe test of the accuracy of the pressure gauge , and the constancy of the records of the thermo-junctions under pressure .
I am not aware of any quantitative experiments on this subject having been made under high pressures .
Sir William Thomson carried the proof of the accuracy of Professor James Thomson 's great theoretical discovery to a pressure of 17 atmospheres.* The experiments of Mousson ( " Pogg .
Annalen , " 1858 ) were not of a quantitative character , being merely intended to show that ice at a temperature of \#151 ; 18 ' C. might still be liquefied by the application of an enormous pressure .
The following experiments appear to show that a convenient manometer for very high pressures , based on the observation of the freezing-point , may be easily constructed .
In all the following experiments the galvanometer , moving to the negative side , represents a cooling effect on the junction inside the * " The Effect of Pressure in Lowering the Freezing-Point of Water experimen- .
tally demonstrated .
" " Phil. Mag. , " 1850 .

