XXIV .
" On the Temperament of Musical Instruments with Fixec Tones .
" By ALEXANDER J. ELLIs , F.R.S. , F.C.P.S.* Received June 8 , 1864 .
In the preceding paper on the Physical Constitution of Musical Chords ( Proceedings , vol. xiii .
p. 392 ) , of which the present is a continuation , I drew attention to the importance of abolishing the distinction between tones which differ by the comma 81 : 80 , on account of the number of fresh relations between chords that would be thus introduced .
The contrivances necessary for this purpose have long been known under the name of Temperament .
I have shown that the musical scale which introduces the comma consists of tones whose pitch is formed from the numbers 1 , 3 , 5 , by multiplying continually by 2 , 3 , and 5 .
Hence to abolish the comma it will be necessary to uise other numbers in place of these .
But this alteration will necessarily change the physical constitution of musical chords , which will now become approximate , instead of exact representatives of qualities of tone with a precisely defined root .
It is also evident that all the conjunct harmonics will be thus rendered pulsative , and that therefore all the concords will be decidedly dissonant at all available pitches .
The result would be intolerable if the beats were rapid .
Temperament , therefore , only becomes possible because very slow beats are not distressing to the ear .
Hence temperament may be defined to consist in slightly altering the perfect ratios of the pitch of the constituents of a chord , for the purpose of increasing the number of relations between chords , and facilitating musical performance and composition by the reduction of the number of tones required for harmonious combinations .
The subject has been frequently treated t , but the laws of beats and composition of tones discovered by Prof. Helmholtz have enabled me to present it in an enitirely new form , and to determiine with some degree of certainty what is the best possible form of temperament .
Let the compound tones P and Q , of which P is the sharper , form the concordant interval p : q. Then P : Q-p .
or qP= pQ , that is , the qth harmonic of P and the pth harmonic of Q are conjunct .
Now let P be changed into P. ( I+t ) , where t is small , and rarely or never exceeds -i=0125 .
Then the qth harmonic of P. ( 1 +t ) will be qP .
( I +t ) and will no longer=pQ .
The difference between the pitch of these harmonics is qP .
( 1 + t)-pQ=qt .
P=pt .
Q. Hence the number of beats in a second produced by this change in P will be found by multiplying the lower pitch Q by pt , which is therefore the beat factor , and will be positive or negative according as the pitch of P is increased or diminished , or the interval is sharpened or flattened .
The other beats which existed between the joint harmonics of the dyad P , Q may be increased or diminished by this change , but in either case so slightly that they may be left out of consideration in comparison with the beats thus introduced .
But the differential tone which was PQ becomes Pt+ PQ , and is therefore a tone which is entirely unrelated to the original chord , and which may be .
come prominently dissonant .
This is an evil which cannot be avoided by any system of temperamenit , and is about equally objectionable in all systems .
It may therefore be also lef out of consideration in selecting a temperament .
The melody will also suffer from the alteration in the perfect ratios .
Any interval is best ineasured by the difference of the tabular logarithms of the pitches of the two tones which form it .
Hence the interval error e=log[P.(lIt ) Q ] -log [ P : Q]=log(I+t)=p t , if the square and higher powers of t be neglected , and , .
be the modulus .
Hence the beat factor which =pt , will =pe* y ?
, or ca pe .
I call pe the beat meter , and represent it by 3 .
We may assume that the dissonance created by temperameit ca 2 Hence for the same just interval p : q , variously represented in different temperaments the dissonance oc e2 , That is , the harmony varies inversely as 132 and the melody varies inversely as e ' .
Hence for the same interval the harmony and melody , both vary inversely as e. The geei'eral harmony and melody may be assumed to be best when 2132 and le ' are minima , which will not happen simultaneously .
The following contractions for the names of the principal intervals will der k. Sichsischen Gesellsehaft der WVissenschaften , vol. iv .
Nachtrige sir Theory der niusikalischen TonvehdIltnisse , ibid. vo .
v. Ueber die wissenschaftliche Bestimmung der musilkalischen Temperatur , Poggendorni 's Annalen , vol. xc .
p. 3S53 .
Araumaant , IUeber die versehiedene Bestimmtng der Tonverhujtnisse umd die Bedeutung des Pythagoreischen oder r einen Quinten-Systems fifr unsere heuLtige Misl , 1858 Bielmnkoltz , Die Lehre von den Toneinpfin dungen , 1803 .
I ama most indebted to Smith , Drobisch , and Helmholtz .
be used throughout this paper .
See also the last columns in Tables XII .
and XIV .
Signl .
Interval .
Examiple .
Signi .
Interval .
Example .
Ist .
U-nison ... . .
cc find .
Major Second cd 2nid .
Minor Second ... .
of IIlrd .
Major Third ... .
ce 3rd .
Minior Third ... .
eg IVth .
Augmiiented Fourthl fb 4-th .
Perfect Fourth .
cf Ath .
Perfect Fifth c gO .
5th .
Dimiiinished Fifth. .
b f2 VIthl .
Major Sixth ca ( I th .
Minor Sixth e c2 VIIIth .
Major Seventh. .
cb 7th .
Minlor Seventh g f2 VIJIVe .
OcttaVe ... ... c IXth .
Major Ninth ... .
c d2 9th .
Minor Ninth efe Xth .
MAajor Tenth .
ce2 10th .
Minor Tenith ... .
e g2 In no system of temperamiient will it be possible to interfere with the octave , the only unisonant concord .
hIence 2 will remain unchanged .
Let the ratios of the tenmpered IIlrd and Vth be T v , which will replace 5 : 4 and 3 : 2 throughout the system of chords .
1-hence if we take four successive perfect major triads in the form 4 , 5,6 as aE G , GB d , df ta , ta c2te2 , and suppose them to be tempered so that the distinction between E and tlE no longer exists , but that in each chordl the pitch of the second and third tones are T and v times that of the first tone respectively , while the ratio of the octave remains unchanged , the ratio of each of the above toiies to C will be as under : c( E , G , B , d , 17 , e , cv , e 7 , TT V ) , 2 TV 2 V3 , TV3 , V4 Hence , since ei=4E , we have v'=4T as the first condition of tempera .
meant , showing that we shall arrive at the same tone whether we take two VhIlves and a tempered Illrd , or take four tempered Vths , as in Ce , cc2 , ci ei , and C G , Gd , dca , aei In this case the above ratios reduce to q , i/ ' ; , c , B , d , f , a , c2 e2 4 , V3 4 , 4 , 3 , 4V 4 .
If we further call the interval of the mean tone in , the lim ma 14 the sharp : , the flat b , and the diesis 0 , the above ratios give dd V " &gt; ' ' 2C- ' G 20 2i c2 , C2 : VI 2 ' =7 C2 ==0 ; : V di &gt; 2V2X ( 24 *v 27 CmV74 V12 Whellee w$1 Ie ?
, a ?
Bi ; Xri14 1y2 .
IHence all intervals and pitches can be expressed in , terms of v. This further appears from arranging the 27 different tones required in .
tempered scales , in order of Vths , thus atbef bbk , ff c'7 , ?
r11 , db~ , ab , eb , 0i , f , c , y , d , t , e , b , f ; ?
, c# , y : , d : , a , , e:+ , b$ , f X , c X , X. It will be obvious from Table V. ( Proceedings , vol. xiii .
faciuig p. 108 ) , when the signs tj are omitted , that these 27 tones suffice for all keys from CO to C &lt; .
This also appears from observing that the complete key of C requires 7 naturals , 3 flats and 3 sharps , or 13 tonles , and that one flat or sharp is introduced for each additional flat or sharp in the signature of the key .
IIence for 7 flats and 7 sharps in the signature 14 additional tones are required , making 27 in all .
The rarity of the modulations into dt , , gb or c ) minor enables us generally to dispense with the three tonles aO , e5b , 60 , anrl thus to reduce all music to 24 tones .
The system of writing music usually adopted is only suitable to such a tempered scale , and therefore requires the addition of the acute and grave signs ( , t ) to adapt it for a representation of the juist scale founded on the numbers 1 , 2 , 3 , 5 .
To calculate the valuie which must be assigned to v so as to fulfil the coniditions supposed to produce the least disagreeable system of temperament , it will be most convenient to use logarithms , and to put log v=log _-x='1760913-r .
The above arrangement of the requisite 27 tones in order of Vths , therefore , enables us to calculate the logarithms of the ratios of the pitches of all the tones to the pitch of c in terms of xz , by colntinual additions and subtractions of log v , rejecting or adding log 2 =-3010300 , when niecessary , to keep all the tones in the same VlIlve .
The result is tabulated in Table XII .
, column T. From this we immediately deduce log mn=log d -log c= '05115262x log I =log f-log e= '0226335+ 5x log ii =logf#-log f= 028.51917t loga =loggb-logf=--O0058851+12x .
To find the interval errors , the just intervals must be taken for the commonest modulations into the subdominanit and dominant keys , as explained in my paper on a Perfect Musical Scale ( Proceedings , vol. xiii .
p. 97 ) .
As the method of determining temperament here supposed makes the errors the same for the same intervals in all keys , that is , makes the temperament equal , it is sufficienlt to determine the interval errors for a single key .
Hence the just intervals are calculated in Table XII .
, column J , for the key of C , and the interval error is given in column c , in terms of x and k=log Ill , the interval of a comrmra .
From these interval errors the beat meters for the six concordant dyads are calculated in column .
To these are added the values of z C2 and 2 j3 ' , also in terms of x and k. If for k we put its value n0053950 , these last expiressions become Ie = 0009314 -1l1437400x+ 420v2 2P*1=000436W59-58158100x+ 199WS2 .
Hence Table XII .
suffices to give complete information respecting the effectof any system of temperament when a is known .
The following are some of the principal conditions on which it has been proposed to found a system of temperament .
I shall first determine the value of x and log v on these conditions , and then compare the results .
A. HARMONIC SYSTEMS OF EQUAL TEMPERAMENT .
I. Systems with two concords perfect .
No. 1 ( 45 ) * .
System of perfect 4ths and Vths .
Here x=0 , log v='1760913 .
This is the old Greek or Pythagorean system of musical tones , more developed in the modern Arabic scale of 17 tones .
No nation using it has shown any appreciation of harmony .
No. 2 ( 2 ) .
System of perfect I1Irds and 6ths .
Ihere e for III , or k-4U-=0 , x-a k=-00134875 , log v=-17474255 .
Helee log m=log d=I0484551_-1 log 5= ( log ?
+1 0 ) , so that the tempered mean tone is an exact mean between the just major and minor tones .
Hence this is known as the System of Mean Tones , or the Mesotonic System , as it will be here termed .
It was the earliest system of temperament , and is claimed by Zarlino and Salinas .
See also Nos. 13 and 19 .
No. 3 ( 23 ) .
System of perfect 3rds and VIths .
Here e for 3 , or -k+3x=0 , xv= k-*0017983 , log v= l 742930 .
IL .
Systems in which h the harmaony of two concords is equal .
No. 4 ( 20 ) .
The IlIrd and Vth to the same bass ; beat equially and in opposite directionst .
Herep for III+ l+ for V=0 , or ( 5k-20.v)-3 =0 =j,5 , , , k 00117251 log v= 1749188 .
No. 5 ( 15 ) .
The 6th and Vth beat equally , and in the same direction+ .
Here , for 6=B for V , or -8k+32x=-3x , x= k8 k=0012331 , log v= 1748582 .
No. 6 ( 21 ) .
The IlIIrd and 4th beat equally , and in the same direction .
Here 3 for III=(3 for 4 , or 5k-20x=4x , x-5k=_0011239 , log v_* 1 749674 .
No. 7 ( 18 ) .
The 6th and 4th beat equally , and in opposite directions .
Here/ 3 for 6+13 for4=0 , or ( -8k+32x)+4x=0 , c=-k=-0011989 , log v=1748924 .
No. 8 ( 16 ) .
The 3rd and Vth beat equally , and in the same direction .
Here / for 3=13 for V , or -6k+18x=-3v , x=T 2x=0015414 , log v=_1745199 .
See No. 20 .
No. 9 ( 13 ) .
The VIth and Vth beat equally , and in opposite directions .
Here P3 for VI+/ 3 for V=0 , or ( 5k15v)-3x=0 , = 58 k= 0014986 , log v=_1745927 .
This coincides with Dr. Smith 's system of equal harmony , as contained in the Table facing p. 224 of his ' Harmonics , ' 2nd ed. No. 10 ( 9 ) .
The 3rd and 4th beat equally , and in opposite directions .
HIere/ for3+pfor4=0 , or(-6k+18 .
) ?
)+4x=O , =-x3 k=-0014713 , log v=-l746200 , No. 11 ( 2 ) .
The VIth and 4th beat equally , and in the same direction .
Here / for VI1=,3 for 4 , or 5k-15x=4x , x=--5k='0014197 , log v=-1746716 .
III .
Systems in which the harmony of two concords is in a given ratio .
No. 12 ( 24 ) .
The beats of the IlIrd and Vth are as 5 : 3 , but in opposite directions .
Here , forIII : : for V=-5 : 3 , or 15k-60x= 15x , x= k='0010790 , log v-'1750123 .
M. Romieu gives this temperament under the title of systbme tempere de 1 comma " M6m .
de l'Acad .
1758 .
See No. 18 .
No. 13 ( 2 ) .
The beats of the 3rd and Vth are as 2 : 1 , and in the same direction .
Here P for 3 : / for V-2 , or -6k+ 18x=-6x , x=1k , as in No. 2 .
No. 14 ( 12 ) .
The beats of the 3rd and Vth are as 5 : 2 , and in the same direction .
Here/ 3for3:3forV5:2 , or-12k+36x=-15 , =x-4 -k=-00126943 log v=-1748219 .
See No. 29 .
IV .
Systems of least harmonic errors .
No. 15 ( 7 ) .
The harmonic errors of all the harmonic intervals conjointly are a minimum .
This is determined by putting the sum of the squares of the beat meters , or ( by Table XII .
) 150k2-1 07Skx+ 1998x2=a minimum , which gives -= k= ; J-00145.54 , log v= 1746359 .
If we had used the sum of the squares of the beat factors , we should have obtained an equation of 16 dimensions inv , which gives logv=-1 746387 .
The difference between the two values of log v is not appreciable to the ear .
No. 16 ( 14 ) .
The harmonic errors of the 3rd , IIIrd , and Vth conjointly are a minimum .
1here ( i for 3)2+(/ 3 for III)2+(13 for V)2 , or ( 6k18 , )2+ ( Sk-20X)2 + 9X2-a minimum , which gives *= 20 / 8= -00 1 5309 , log v= 1 744404 .
No. 17 ( 6 ) .
The harmonic errors of the Vth and IlIrd conijoinitly are a minimuim .
Here ( / for Y)'+ ( / for IT )2 , or 9x2+ ( 5k-20x)2=a minimum , x=1109 k=_0013190 , log v=-1 4723 .
B. MELODIC SYSTEMS OF EQCUAL TEMPERAMENT .
V. Systems of equtal or eqzal and opposite interval errors .
No. 18 ( 24 ) .
The interval errors of the ITIrd and Vth are equal and opposite .
HIere e for 111+ e for V=-0 , or k4 x= x , x le , as in No. 12 .
No. 19 ( 2 ) .
The interval errors of the 3rd and Vth are equal .
Here c for 3 =e for V , or -lc+33x=-x x-4Ik as in No. 2 , No. 20 ( 16 ) .
Tle interval errors of the IlIrd and 3rd are equal .
Here e for IlJ=e for 3 , or k-4 , x=-k+3x , x=-k , as in No. 8 .
VI .
Systems in which the interval errors of two intervals are it ?
a given 2ratio .
No. 21 ( 17 ) .
The errors of the IlIrd and Vth are as 5 : 3 , but in opposite directio'ns .
Here e for III : e forV =-5 : 3 , or 3k-l12xr=5x , x=-IL k-x '0015750 , log v= 1745163 .
This is the theoretical determination of M. Romieu 's anacratic temperamenit ( Meum .
de l'Acad .
1758 , p. 510 ) , to which , however , he has in practice preferred No. 22 .
No. 22 ( 29 ) .
The errors of the ITIrd and Vth are as 2 : 1 , but in opposite directions .
Here e for III : e for V=-2 , or kc-4x=2x , x= k='0008975 , log , v=1 7 17938 .
This is M. Romieu 's anacratic temperament .
See No. 21 .
No. 23 ( 26 ) .
The errors of the IIIrd and Vth are as 194:1 , and in opposite directions .
Here e for III : e for V=194 , or k-4-=1'94.r , ?
, = 0 k-*0009683 , log v= 1751830 .
This is the temperament calculated by Drobisch ( Nachtriie , ? ?
10 ) from Delezenne 's conclusion ( Rec .
Soc. Lille , 1826-27 , pp. 9 and 10 ) , that the ear can detect an error of -284k in the IlIrd , and *146k in the Vth , which gives the comparative sensibility as *284 : 4 146 =I 94 .
No. 24 ( 20 ) .
The errors of the IIIrd and 3rd are as 2 : 5 , but in opposite directions .
HIere e for III for e 3-2 : 5 , or 5k1-20x =2k-6x w=T-13 k= 0011561 , log v=*1749352 .
See No. 27 .
No. 25 ( 46 ) .
The errors of the 3rd and lIrd are as 2 : 1 , but in opposite directions , or the errors of the Vth and 3rd are equal and opposite .
Here e for 3 : e for III= -2 , or 2k-6x=k-4x , or else x= -k+3xX both give x=2 k= 002697.5 , log v=K1733938 .
Here the error of the Vth reaches the utmost limit of endurance .
VII .
Systern of least melodic errors .
No. !
-26 ( 1 ) .
The interval errors of all the melodic intervals conjointly are -C .
COMBINED SYSTEMS OF EQUAL TEMPERAMENT .
No. 30 ( 4 ) .
The combined harmonic and melodic errors are a minimum .
By combining the equations of No. 15 and No. 26 , we have ( 539 + 106 ) kC ( 1 998 +420)x , or x= A44a58 J= 001444 , log v=1 746439 .
No. 31 ( 32 ) .
The tones are a mean between those of No. 1 and No. 2 .
Here x=(sum of the two values of x in No. 1 and No. 2 ) =*0006744 , log v= 1754169 .
This is proposed by Drobisch ( Nachtrage , ? ?
9 ) .
No. 32 ( 42 ) .
The errors occasioned by using the tempered c , d , f , , g , 6bb , c2 &gt; , c2 for the just c , d , e , fa g , a , 6 , C2 are a minimum .
Using s for *0004901 , and forming the values of these errors by Table XIT .
, we have4X2 +2 ( -S X)2 ? ?
2+ ( s-9X)2 ?
( s7x)'=a minimum , x--24A S-=000059084 , log v= 1760322 .
This is proposed by Drobisch as a system of teniperament adapted to bowed instruments ( Mus. Tonbestim . ? ?
57 ) , allowing them to use a system of perfect fifths , and yet play the perfect scale very nearly by substitution .
Such a system would be more complicated than the just scale for any instrument , and would require maiy more than 27 tones .
It is , therefore , unnecessary for the violin , and impossible on instruments with fixed tones .
D. CYCLIC SYSTEMS OF EQUAL TEMPERAMENT .
When it was supposed that the number of just tones required would be infinite , importance was attached to cycles of tones which by a limited number expressed all possible tones .
Hence Huyghens 's celebrated Cyclus Harmonicus , which he proposed to employ for an instrument with 31 strings , struck by levers and acted upon by a moveable finger-board ( abacus mobilis ) , acting like a shifting piano or harmonium .
The condition of forming a cycle is not properly harmonic or melodic ; it is rather arithmetic .
If log v : log 2 be converted into a continued fraction for any of the preceding values of log v , and y : z be any of the convergents , then , putting log 2=z .
h3 we shall have log v=yh , which is commensurable with log 2 , and consequelntly the logarithms of all the intervals will be multiples of h , and therefore commensurable with log 2 .
A cycle of z tones to the octave will thus be formed .
If z is less than 27 , the number of tones otherwise necessary , the cycle may be useful , otherwise it can only be judged by its merits as an equal temperament .
As an historical interest attaches to several of these cycles , I subjoin a new method for deducing them all , without reference to previous calculations of log v. Since log v=y .
h , and log =7 log 2-12 log v= ( 7z-12y ) .
h , we have only to put 7z -12y= ... -2,1 , 0 , 1 , 2 , ... and find all the positive integral solutions of the resulting equations .
This gives for 7z-12y=-2 , :il3 27 41 55 69 z ; ~ 22 " ' 46 j ' TO$ 'i IS ' .
7z-12y=1 , Y_3 10 17 24 31 38 45 52 7z-12y= 0 , y_7 7Z-12y 1 , y4 11 18 25 32 39 46 53 z 7 ` 19 ' 31 ' 43 ' 55 ' 67 ' 79 ' 9l ' * 7z12y= 2 , y_1 15 29 43 57 7z-12y= 3 , y-5 12 19 26 33 40 47 54 z 9 ' 21 ' 33 ' 4-5 ' 57 ' 69 ' 81 ' 93'~~ Many of these cycles are quite useless .
The following selection is arranged in order of magnitude , from the greatest to the smallest cycle .
No. 33 ( 38 ) .
Cycle of 118 ; h=-0025511 , logv=69h= 1760259 .
This is Drobisch 's cycle ( Mus. Ton . ? ?
58 ) representing No. 32 .
No. 34 ( 8 ) .
Cycle of 93 ; h=-0032368 , logv=54ht=-*1747872 .
This may represent No. 2 .
No. 35 ( 3 ) .
Cycle of 81 ; h=*0037164 , log v=47h=*1746708 .
This may represent No. 11 ( 2 ) .
No. 36 ( 39 ) .
Cycle of 77 ; h=-0039095 , logv=45h=-1759275 .
This is the same as No. 52 .
No. 37 ( 19 ) .
Cycle of 74 ; h= 004068 , logv=43h=-*1749200 .
This is another of Drobisch 's cycles ( Nachtrige , ? ?
7 ) representing No. 27 .
No. 38 ( 22 ) .
Cycle of 69 ; At=*004363 , logv=40h_=1745200 .
No. 39 ( 28 ) .
Cycle of 67 ; h=-004493 , logv=39h='1752270 .
No. 40 ( 40 ) .
Cycle of 65 ; h=-0046123 , logv=38h=_17598674 .
No. 41 ( 27 ) .
Cycle of 57 ; h='0052812 , logv=33h= 1742796 .
No. 42 ( 30 ) .
Cycle of 55 ; A=-0054733 , logv=32h=_1751456 .
This is mentioned by Sauveur ( Ma'am .
de l'Acad .
1707 ) as the commonly received cycle in his time .
Esteve ( loc. cit. p. 135 ) calls it the Musicians ' Cycle .
No. 43 ( 11 ) .
Cycle of 50 ; h=-0060206 , logv=29h='l745974 .
This is Henfling 's cycle ( loc. cit. p. 281 ) , and is used by Dr. Smith to represent No. 9 .
No. 44 ( 43 ) .
Cycle of 53 ; h_=0056798 , log v=31h= 1760800 .
This is the cycle employed by Nicholas Mercator ( as reported by Holder , 'Treatise on Harmony , ' p. 79 ) to represent approximately the just scale .
He did not propose it as a system of temperament as has been recently done by Drobisch ( Musik .
Tonbestim .
Einleit .
) .
It was the foundation of the division inito degrees and sixteenths adopted in my previous paper Proceedings , vol. xiii .
p. 96 .
No. 45 ( 37 ) .
Cycle of 45 ; h=-006689 , log , v=26h1=* 1738940 .
No. 46 ( 25 ) .
Cycle of 43 ; h=-0070007 , log v=25hA= l750l75 .
This is Sauveur 's cycle , defended in Men .
de I'Acad .
for 1701 , 1702 , 1707 , and 1711 .
No. 47 ( 10 ) .
Cycle of 31 ; h= 009711 , 1og v=i18=-11747900 .
This is Huyghens 's Cyclus Harmonicus , which nearly represents No. 2 ( 2 ) .
It was adopted , apparently without acknowledgment , by Galin ( Delezenne , loc. cit. p. 19 ) .
No. 48 ( 44 ) .
Cycle of 26 ; h-=011578 , log v= 1h=-1736700 .
No. 49 ( 25 ) .
Cycle of 19 ; h=70158437 , logv=11h= 1742807This is the cycle adopted by Mr. Woolhouse ( Essay on Beats , p. 50 ) as most convenient for organs and pianos .
It may therefore go by his liarne , although it is frequently mentioned by older writers .
It is almost exactly the same as No. 3 ( 23 ) .
No. 50 ( 35 ) .
Cycle of 12 ; h=*0250858 , log v=7h-=1756008 .
As this is a cycle of twelve equal semitones , it may be termed the Hemztonic temperament .
It is the one most advocated at the present day , and generally spoken of as " equal temperament " without any qualification , as if there were no other .
It was consequently referred to by that name only in my former paper ( Proceedings , vol. xiii .
p. 95 ) .
For its harmonic character see No. 53 .
E. DEFECTIVE SYSTEMS OF EQUAL TEMPERAMENT .
It has been from the earliest times customary to have only twelve fixed tones to the octave , on the organ , harpsichord , piano , &amp;c. , and to play the other fifteen by substitution , as shown below , where the tones tuned , arranged in dominative order , occupy the middle line , and the tones for which they are used as substitutes are placed in the outer lines , and are bracketed .
[ ,40 , E- , O , -Mb , Fb , cb , Ob D7 , wA5 ] b , B &gt; , F , C , G , D , A , E , B , F , C C ?
, G : ED , A , E ?
, B# , F x , C x , G X ] The consequence was , that while the Vths in the middle line were uiniform , the Yths and 4ths produced in passing from one line to the other ( as G:E1 for AbEb or G:D : ) were strikingly different .
Similar errors arose in the other concordant intervals .
It is evident that the interval error thus produced must be the usual interval error of the system ilnereased or diminished by the logarithm of the diesis , where log =logg -logf = -'0058851+12x=-kc-s+12x , where s=-0004901* .
Such interval errors are termed wolves , from their howling discordanice .
In Table XIII .
will he found an enumneration of all the wolves , with a notation for them , and an expression of their interval errors and beat meters in terms of k , s , and x. No. 51 ( 33 ) .
System of least wolf melodic errors , The sum of the squares of the wolf interval errors , or 2k12+2ks+6s2-4 ( 1k/ +28s ) x+266X2 , isa minimum .
Hence 22k+56s=266i , or x=0005495 , logv='1755418 .
No. 52 ( 39 ) .
System of least wolf harmonic errors .
The sum of the squares of the wolf beat meters , or 25k/ 2 +1 75s + 50ks(5.5Ok+3072s ) x+ 13662x ' , is a minimum .
Hence 275k+1536s_=13662x , or x_=0001638 , log v =1759275 , as in No. 36 ( 39 ) .
No. 53 ( 35 ) .
The wolf initerval errors are equal to the usual interval errors , that is , there are no wolves , or there are none but wolves .
In this case log B=0 , or , since =gb 2:f:27 2v , 7 log 2 =12 log v. Ihence this systenm is the cycle of 10 , No. 50 .
When a is greater than 1 , Yg is sharper thanf , and log v is less than 172 log 2 , or1 1756008 .
But if B is less than 1 , g7 isflattean thanf athan log 2 , or *1756008 .
The latter case is , according to Drobiseb , indispensable for musical theory and violin practice ( MIusik .
Tonbestim .
Einleit .
) .
Since this temperamrient thus forms the boundary of the two other classes , distinguished by yb being flatter or sharper than f# , Drobisch terms it the " c meani " temperament ( ibid. ? ?
51 ) .
It is this property of making gy =f : which renders this temperament so popular , as the ear is never distressed by the occurrence of intervals different from those expected , and the whole number of tones is reduced to 12 .
No. 54 ( 31 ) .
The wolf interval error of the IlIrd is to its usual interval error as 14 : 5 .
This gives -s+8y : kc-4x14 : 5 , or 96x=14k+5s , x='0008123 , log v= '1752790 .
This is Marsh 's system of temperament ; see Phil. Mag. vol. xxxvi .
p. 437 , and p. 39 seqq .
Schol .
8 .
No. 55 ( 36 ) .
The wolf errors of the IIlrd and Vth conjointl.y are a minimum .
Here ( -s+ 8x)2J+(-k5s+ II )2 iS a minimum , whence l1k+19s =185xx , X=*00037121S , l ogv='1757201 .
No. 56 ( 37 ) .
The wolf errors of the Vth and IlIrd are equal and opposite .
fIere-k-s8+llx=s-8x , 19x= =+2s , x='0003356 , logv=_1757557 .
No. 57 ( 34 ) .
There is no Vth wolf .
Here k-s +1 10 , = '0005351 , log v= ' 1755562 .
No. 58 ( 41 ) .
There is no llrd wolf .
Here -8+8x=40 , x=*0000613 , log v= 1760300 .
This is almost exactly No. 32 ( 42 ) .
No. 59 ( 42 ) .
There is no 3rd wolf .
Here rv-9x==0 , x= 0000545 , log v=_ 1760368 .
F. SYSTEMS OF UNEQUAL T.EMPERAMENT .
In a defective equal temperament the same just concordance is represented by two different discordances .
As performers limited themselves to twelve tones to the octave , those who found the Hemitonic temperament No. 50 ( 35 ) too rough , accepted this variety of representatives of the same con- ' cordance as the basis of a temperament , hoping to have better IIIrds in the usual chords , without the wolves of the defective temperament .
Others conceived that an advantage would be gained by altering the character of the different keys .
Thus arose utnequal temperament , properly so called , which must be carefully distinguished from any defective equal temperament with which it is popularly confused .
Arrange the twelve unequally tempered chords as follows , where the identical numbers indicate identical chords with different names 1 .
CE G. 7 .
FJ J# cj , .
7 .
G Bb dX .
2 .
GB d. 8 .
C : E : G# .
8 .
Db F A &gt; .
3 .
D FJ S. 9 .
G0 B , d# .
9 .
Ab c eb .
4 .
A eC e. I0 .
D : Fx t ; .
10 .
EG B &gt; .
5 .
E G.BE .
11 .
A cx e ? .
11 .
Bb d. 6 .
Bd f# .
12 .
E : , Gx B$ .
12 .
FA c. Let T+ , t1 , v , be the ratios of the IJIrd , 3rd , and Vth in the nth chord , so that , for example , in the 6th chord d:=T0 .
B , t d. , fJ=v6 .
B. Then it is evident from the above scheme that there exist 12 pairs of equations between these 36 ratios , of the form T11 .
t1 , V ?
, and 4T , =Vn * Un .1 * Un +2 *.n3 ( where , whenl the subscript numbers exceed 12 , they must be diminished by 12 ) , and one condition , VI * V2 * V3 .
V4 o1+Y5+Y9=Y2+Y6+YJoo=3+Y7+Yl 1=40103000+ ... .(a ) x5-$X1 + ?
, JY2 , 'Oo = $i+ ?
/1O- ?
/9 ) x62 + Y2-J3)io 10=X2 +Y11-Y1o0 X7 = '3 + Y3YO 11= X3 + Y2-Yll , tV8=X4+Y4-_y5 X12=-i+Yl Y12 A system of unequal temperament may therefore be determined by arbitrarily selecting eleven different Vths , or else eight different IlIrds and three Vths .
The equations ( a ) show that if the temperament is not equal ( in which case all the y 's are equal , and the interval error of the IlIrd is Xx *0103000=-0034333 , as in the Hemitonic temperament ) , at least four IlIrds must have their interval errors greater than 3x *0103000 , that is , there must be at least four Illrds in every unequal temperament which are inferior to the very bad IlIrds of the Hemitonic system .
Kirnberger , Dr. T. Young , and Lord Stanhope* , in the unequal systems they propose , have each seven IlIrds sharper , and therefore worse than the Hemitonic IlIrds .
In one of Prof. de Morgan 's unequal temperaments , six IIlrds are sharper than the Hemitonic ; in another four are sharper and four the same ; in a third all are the same , but the Vths differ t. Hence nothing is gained over the Hemitonic system in the way of harmony , while the uniformity in the representation of the uniformity of just intonation is entirely lost .
In selecting a temperament , therefore , we may dismiss all unequal temperaments , as they must be inferior to the Hemitonic in both harmony and melody , and will have no advantage over it in the relations of chords or the number of tolnes required .
Also , if it is considered niecessary to play in all keys with only twelve tones , any system of defective equal temperament will be inferior to the I-enlitonic , on accounlt of the various and distressing wolf intervals which occur when the music is not confined to the six major scales of 1t ' , F , C , G , D , A , and the three minor scales of g , d , a. Hence the two conditions of having only twelve tones ( exclusive of octaves ) and of playing in all keys , at once exclude all temperaments but the Ilemitonic .
As , however , organs , harpsichords , and pianofortes with 14 , 16 , 17 , 19 , 21 , 22 , and 24 tones to the octave have been .
actually coristructed and used T , as Mr. Liston used 59 ?
, Mr. Poole used 50 , and Gen. T. Perronet Thomp* Kirnberyer , Kunst des reinieni Satzes in der Musik .
Dr. T. Young , loc. cit. Charl es Earl Stanihope , Principles of the Scienlce of Tun-ing , 1806 .
t -De Morgan , loc. cit. p. 129 , temperamtients Q , R , S. v Mr. Farey ( Phil. Nag .
vol. xxxix .
p. 410 ) gives the particulars of their scales , builders , and localities . ?
The following ) account of Alr .
Liston 's organ is deduced from the data of Mr. Farey ( Phil. Mag , .
vol. xxxix .
p. 418 ) .
Scale : c lc tc4 c9 d9 1c td2l TCX Itd cx d td td# e5 d# te5 tte5 of te tfP tt , fb 4I ev f te : tf tf .
f ?
t -1f4 tg( i ?
fx 4g fx y tg 4:g taJ , t tta a ta bOb tOb ta$ a# bb J-a$ tbO lb b c0 tb tcd lb# tc b# .
Chords : Table V. col .
III .
, lines 4 to 13 ; son now uses 40 tones to the octave on their justly intoned organs , the conidition of having twelve tones and lno more , does liot seem to be inievitable .
It will therefore be necessary to determiine what would be the best system of temperament for the complete equally tempered scale of 27 tonles , and how great a sacrifice of musical effect is required by the use of the Hemitonic system .
In Table XV .
I have calculated for each of the 59 ( reducing to 51 ) systems of equal temperament already lnamed , the interval errors of the Vths , IIIrds , and 3rds , and the sums of the squares of the 23 initerval errors and the 6 beat meters of Table XII .
I have then arranged the temperaments in order according to each of these five results , and number ed the order .
Fiinally , I have added the five order numbers together and arranged the whole in the order of these sums .
The smallest number would therefore clearly give the best temperament , supposing that all the five points of comparison were of equal value .
Now the first and second temperament on the list , or No. 26 and No. 2 , only differ from each other in the fifth , sixth , or seventh place of decimiials with respect to these five results , a difference which no humani ear , however finely constituted by nature or assisted by art , could be taught to detect .
As No. 2 , or the Mesotonic system , is deter minled in the simplest mainner , I consider it as the real head of the list .
There is , however , little to choose betweeni it and any one of the ten or twelve systems which follow , except in sinmplicity of conistruction and comparative ease in realizationi .
The Hemitonic system , however , comes 35th irL the list , and the old Pythagorean , recently defended by Drobisch and Naumann ( op .
cit. ) , and asserted to be the system actually used by violinists , is the 45th .
No onie who has heard any harmonies played on the Pythagorean system will dispute the correctness of the position here assigned to it , which fully explains the absence of all feeling for harmony among the niationls which use it the ancient and modern Greeks , the old Chinese , the Gaels , the Arabs , Persians , and Turks .
No modern quartett players could be listened to who adopted it .
The contest lies , therefore , between the Mesotonic and the Hemitoniic systems .
The Mesotoliic is that known as " the old organ-tuning , " or , sirnce it was generally used as a defective twelve-toned system , as the " unequal temperament .
" Within the limits of the nine scales already named , the superiority of the Mesotonic to the Hemitoniic system has long beeni practically acknowledged .
But the extremely disag , reeable effect of the wolves ( more especially to the performer himself ) has finally expelled the system from Gernmany altogether , and from Enrgland in great measure .
On the pianoforte IV .
, 5 to 14 ; V. , 5 to 16 ; AI .
, 6 to 15 ; VII .
, 7 to 16 ; VIII .
, 7 to 17 ; IX .
, 9 to 1.3 ; X. , 9 to 13 .
Tones not forming part of any chord and required chiefly by the system of tuning : t(t tte5 tfP ttfp tg5 tta b7 tbb tfc .
Complete keys : F , C , G , D , tA ; E , B , Fo .
The keys of Xi , Bb had their synonymoius , and tE , tB their relative minors perfect .
the Hemitonic system is universally adopted in intention .
It is , however , so difficult to realize by the ordinary methods of tuning , that " equal temperament , " as the Hemitoinic system is usually called , has probably never been attained in this country , with any approach to mathematical precision .
In Table XIV .
I have given a detailed comparison of the Mesotonic atid Hemitonic temperaments with each other and with just intonation , for the system of C ( Proceedings , vol. xiii .
p. 98 ) , from which the great superiority of the Mesotonic over the Hemitonic both in melody and harmony becomes apparent .
But this comparison rests upon the preceding calculations , which were founded upon the beats that arise from rendering the conjunct barmonics pulsative .
It was therefore assumed that the qualities of tone employed were such as to develope these beats .
The result will consequently be materially modified when the requisite harmonics either do not exist or are very faint .
Now for the Vth the conjunct harmonics are 2 and 3 , 4th , , 3 and 4 , VIth , , 3 and 5 , IlIrd , , , , 4 and 5 , 3rd , , 5 and 6 , 6th , , 5 and 8 .
If then only simple tones are used , as in the wide covered pipes of organs , or such quialities as develope the second harmonic only , such as tuning-forks , to which we may add flutes , which have almost simple tones , iio beats will be heard , and any system of temperament may be used in which the ear can tolerate the interval errors .
Now Delezenne 's experiments show ( loc. cit. ) that a good ear distinguishes in the unison an interval error of 0-2807k , VlIIve , , , , 031k , Vth , , , , 01461k , IIlrd , , , , 0284k , VIth , , , , 0299k , and an indifferent ear perceives an error of 0f561k in the VIlIve , and 0292k in the Vth .
We may say , therefore , generally that the ear just perceives an interval error of 4k in the Vth , and Ak in the other intervals .
Now in the Mesotonic system the interval error of the Vth is -'l , and therefore just perceptible , but in scarcely any other interval does it exceed +k .
Thus it is -Ik in the VIlth , 0 in the IIlrd , and +k in the Vlth , and it is therefore in those intervals imperceptible .
In the Hemitonic system the error of the Vth is --1Jk , and hence quite imperceptible , but the errors of the Vllth , Illrd , and VIth are respectivelyT-1Lk , iTk , and -1a-k and therefore perfectly appreciable .
It is only in the VlIth that this error is at all agreeable .
The sharpness of the IIlrd and Vith is universally disliked .
Hence in those qualities of tone which are most favourable to the Hemitonic system , it is much inferior to the Mesotonic .
In , Table XV , the Mesotonic stands 2nd in order of melody , inappreciably different from the 1st , and the iemitonic 39th .
If the 3rd harmonic only is developed in the qualities of tone combined , the beats of the Vth are heard , but those of the other intervals are not perceived .
The beats of the IlIrd and VIth , which are so faulty on the Hemitonic system , will not be perceived at all unless the 5th harmonic be developed , and will not be much perceived unless it be strongly developed .
Now the 5th harmonic is comparatively weak on all organ pipes and on pianofortes , and hence the errors are not so violently offensive on these instruments .
If , however , the 'mixture stops , ' which strengthen the upper harmlonics by additional pipes , are employed on the organ , the effect is unmistakeably bad , unless drowned by din or dimmed by distance .
On the pianoforte , however , these intervals , and even the still worse 3rd and 6th , depending on the 6th and 8th harmonics , which are undeveloped on pianoforte strings , are quite endurable .
Hence the Hemitonic system , except as regards melody , will not be greatly inferior to the Mesotonic on a pianoforte and on soft stops of organs , but will only become offensive on loud stops .
But for harmoniums and concertinas , violins and voices , where harmonics up to the Sth , and even higher , are well developed , the Hemitonic temperament is offelsive .
The roughness of harmonimns is almost entirely due to this mode of tuning .
The beats of the VIth , IlIrd , and 3rd are distinctly heard , and the development of differential tones is so strong as frequently to form an unintelligibly inharmonious accompaniment* .
Conicertinas having 14 tones to the octave are indeed generally tuned mesotonically ( or intentionally so ) , thus c c , d d : , e e5 , ff# , g g# , a ab , b b6 .
They are , however , occasionally tuned hemitonically ( or intentionally so ) to accompany pianofortes , thus c c , d d ; e dB , ff : , g yg , a qi , , 6 a : .
Hence it is easy to compare the different effects of the two systems as applied to the same quality of tolne , for harmonies which are common to both .
Having two concertinas so tuned , and a third tuned to just intervals , I have been able to make this comparison , and my own feeling is that the Mesotonic is but slightly , though unmistakeably , inferior to the Just , and greatly superior to the Hemitonic .
There are two other points in which the complete Mesotonic system possesses advantages over the iemitonic .
The Mesotonic VIIth is rather flat , but by using the flat VlIlth in its place , when the harmony will allow , the effect of an extremely sharp VIIth is produced , which is sometimes desirable in melodies .
Thus log Mesotonic VIII &gt; = 28195 , which is sharper even than log Pythagorean VII -27840 .
The ordinary and flatter VIIth can be used when necessary for the harnmony .
Again , by using the German sharp VIth in place of the dominant 7th , that is , by using the chords d Bi DI e , 191 , FA1 , b , A , ) CEb f : , -9 &gt; G B &gt; c , B &gt; D Fg , FA C d ; , CEG a ; , GBD et , ,DF A # , A Of x , E G( Bc x , B DI F : g x , in place of G &gt; B5DJ ?
P , 19 F At c ) , A &gt; Eb g &gt; , -2 G BP clt , B &gt; DF a &gt; , FAC eb , CEG bb , GB f , ID FA c , A C# Eg , E G# B d , B D9 F$ a , when the progression of parts will allow , an almost perfect natural seveinth , better than that obtained by using the corresponding just tones , will result , producing beautiful harmony ; for log Mesotonic VI = 24228 , log=-124304 , and log Just VI : =-24497 .
The ordinary sharper 7th can be used when necessary .
Neither of these effective substitutions is possible on the Hemitonic system .
Considering that singers and violinists naturally intone justly ( Delezenne , toc .
cit. ) , and that the interval errors of the Mesotonic system seldom exceed the natural errors of intonation which may be expected from the inability of the ear* to appreciate minute distinctions of pitch , it appears desirable to tune harmoniums at least , and perhaps organs , mesotonically .
Except as an instrument for practising singers , however ( for which purpose it would be superseded by a Mesotonic harmonium ) , it would be unnecessary to alter the Hemitonic tuning and arrangement of the piano .
But it would be best to teach the Mesotonic intonation on the violin in preference to the Hemitonic , as proposed by Spohr * .
As , however , it would be useless to tune mesotonically with only 12 tones to the octave , it is necessary to have some practical arrangement for 27 , 24 , or 21 tones at least .
I propose the following plan for 24 tones , and as these are exactly twice as many as on pianos , &amp;c. of the usual construction , I call my arranigement the DUPLEX FINGER-BOARD .
PIcan .
cx fx.x xIbI HI l11L0 t~~~ I. ' .
B..f .
DBC L ) EFIA IB D|M Eront ]Yevat9o.Z YeYlalRed Let the black and white manuals remain as at present , and let a yellow manual , of the same form as the black , be introduced between B and C , and E and F. Cut out about the middle third of each black and yellow manual , up to half its width , on the right side only , and introduce a thin red manual rising as high above the black or yellow as these do above the white .
Over G , A , and D , each of which lies between two black manuals , introduce three yellow metal manuals ( lacquered or aluminium-bronze ) shaped like flute keys , and standing at the height of a red manual above the white one , which can therefore , when necessary , be reached below it .
The 7 white manuals are the 7 naturals ; the 5 black manuals are the.5 usual sharps , c d~ fa 9~ aS ; the 2 long yellow manuals are the unusual sharps e~ 6b , and the 3 metal yellow manuals are the double sharps fx gxc x ; and the 7 thin red manuals are the 7 flats , cL , d5 , e , , g !
ab , 0 b7 ' .
The shapes of the red and metal manuals were suggested by those of General T. Perronet Thompson 's quarril8 and Yutal8 .
The 24 levers opening the valves on the organ or harmonium would lie side by side , being made half the width of those now in use , and metallic , if required for strength .
The organ pipes or harmonium reeds would be arranged in two ranks of 12 for each octave , the first rank containing the 7 naturals and 5 usual sharps , and the back rank containing the 7 flats , 2 unusual and 3 double sharps .
The use of this finger-board is accurately pointed out by the ordinary musical notation which distinguishes the sharps from the flats , and is therefore in no respect adapted to the Hemitonic fusion of sharps and flats into mean semitones .

