1895 .
] Mr. A. McAulay .
Octoniom .
169 Arutodesmus , which suggests this link , is at present placed in the Procolophonia , a group separated from its recent association with Pareiasaurus and restored to its original independence , because it has two occipital condyles , with the occipital plate vertical , and without lateral vacuities , and has the shoulder girdle distinct from Pareia-sauria in the separate pre-coracoid extending in advance of the scapula .
III .
" Octonions .
' ' By Alex .
McAulay , M.A. , Lecturer in Mathematics and Physics , University of Tasmania .
Communicated by Rev. N. M. Ferrers , D.D. , F.R.S. Received November 28 , 1895 .
( Abstract .
) Octonions is a name adopted for various reasons in place of Clifford 's Bi-quaternions .
Formal quaternions are symbols which formally obey all the laws of the quaternion symbols , q ( quaternion ) , x ( scalar ) , p ( vector ) 0 ( linear function in both its ordinary meanings ) , 0 ' ( conjugate of 0 ) , i , j , Jc , \#163 ; , Ky , S q , T q , U q , Vq .
Octonions are in this sense formal quaternions .
Each octonion symbol , however , requires for its specification just double the number of scalars required for the corresponding quaternion symbol .
Thus , of every quaternion formula involving the above symbols there is a geometrical interpretation more general than the ordinary quaternion one , an octonion interpretation .
The new interpretation , like the old , treats space impartially , i.e. , it has no special reference to an arbitrarily chosen origin or system of axes .
If Q is an octonion and q a quaternion , the symbols , which in octonions correspond^ Kg , S q , T q , Ug , Yq in quaternions , are denoted by KQ , SQ , TQ , UQ , MQ .
KQ is called the conjugate of Q , SQ the scalar octonion part , TQ the augmenter , UQ the twister , and MQ the motor part .
If A is a " motor " whose axis intersects the axis of Q perpendicularly , QA is also a motor intersecting Q perpendicularly .
UQ .
A is obtained from A by a combined translation along and rotation about the axis of Q ; i.e. , by a " twist " about the axis of Q. TQ .
A is obtained from A by increasing the " rotor " part of A in a definite ratio , and by increasing the " pitch " of A by a definite addition .
UQ = UiQ U3Q , where when Q is thus regarded as an operator , UjQ is a " versor , " i.e. , it effects the rotation mentioned ; and U3Q is a " translator , " i.e. , it effects the translation mentioned .
Similarly , TQ = rI\Q .
T3Q where TiQ is a " tensor , " i.e. , it effects the ratioVOL .
LIX .
n 170 Mr. A. McAulay .
[ Dec. 12 , increase mentioned ; and T2Q is an " additor , " i.e. , it effects the addition-increase mentioned .
The amount by which the pitch of the motor operand A is increased is called the pitch of the octonion operator Q. When a motor B is itself thus considered as an operator its versor is a quadrantal versor and its translator is unity , it does not translate the operand at all ; hence the twister of a motor is a quadrantal versor .
The tensor of B is the magnitude of its rotor part , and the amount by which it increases the pitch of the operand is its own pitch .
These results are all established by aid of quaternions on a purely Euclidean basis .
They have , of course , mechanical interpretations in connection with ( 1 ) the instantaneous motion of a rigid body , ( 2 ) a system of forces , ( 3 ) the momentum of a system of moving matter , and ( 4 ) a system of impulses .
The corresponding motors are called velocity motors ( Sir Robert Ball 's " twist on a screw " ) , force motors ( " wrench on a screw " ) , momentum motors , and impulse motors ( " impulsive wrench " ) .
If A and B are two force motors A + B is the force motor of the system of forces obtained by the composition of the two systems corresponding to A and B ; and similarly for the motors of the other types .
The octonion operator Q( )Q-1 is exactly analogous to the qua-tericn operator q( )q~1- It displaces the octonion operand in the most general manner as a rigid body ; it translates the operand parallel to the axis of Q through a distance double of that through which Q when regarded as a motor operator translates the motor ; and it rotates it as a rigid body round the axis of Q through double the angle of Q. If A is the velocity motor of a rigid body which in the time t has suffered the displacement Q( )Q-1\gt ; A = 2MQQ-1 , and if R is any octonion fixed relative to the body , R = MAMR .
The geometrical connections between A , B ( two motors ) , MAB , and SAB are examined .
The axis of MAB is the shortest distance between A and B. Its rotor part bears the same relation to the rotor parts of A and B that the vector V\#171 ; /3 does to the vectors \#187 ; and / 3 .
The pitch of MAB is the sum of d cot 9 and the pitches of A and B , where d is the distance and 0 the angle between A and B , d being reckoned positive or negative , according as the shortest twist which will bring the rotor of either A or B into coincidence , both as to axis and sense , with the rotor of the other is a right-handed or left-handed one .
A scalar octonion such as SAB requires two ordinary scalars , SiAB and sAB to specify it .
S , AB bears to the rotor parts of A and B the Octonions .
1895 .
] same relation as the scalar S xftbears to the vectors as and sAB is most simply described mechanically .
If A is the velocity motor of a rigid body on which the force motor B is acting , \#151 ; sAB is the rate at which the corresponding system of forces is doing work on the body , _sAB is , therefore , the product of the tensors of A and B , multiplied by what Sir Robert Ball calls the virtual coefficient of the two corresponding screws .
If A and B are rotors , sAB is + six times the volume of the tetrahedron which has A and B for a pair of opposite edges .
sAB/ S^B is what is called the pitch of the scalar octonion SAB .
It is the sum of \#151 ; tan 6 and the pitches of A and B. If , A , B , C are three motors , StA .
BC bears to their rotor parts the same relation that the scalar Sxft-/ bears to the three vectors x , ft , 7 .
The pitch of SABC is the sum of cot tan 0 , and the pitches of A , B , and C ; where d and 0 are related to A and B , as before , and where e is the distance , and 0 the angle between C and the shortest distance of A and B. The rotor part M , ABC of the motor MABC bears , as to direction and magnitude the same relation to the rotor parts of A , B , C as the vector Vac ft-/ bears to the vectors ft , 7 .
The pitch of MABC is the sum of ( e tan 0\#151 ; deot #)/ ( cot2 6 tan2 0 + cot2# + tan2 0 ) and the pitches of A , B , and C. Similarly as to the rotor M].(MAB)C of the motor M(MAB)C .
The pitch of M(MAB)C is the sum of d cot + e cob 0 and the pitches of A , B , 0 .
A finite motor , whose pitch is infinite , is called a lator .
[ The term " vector " is not here used , because though lators and vectors have the same fundamental geometrical properties , and obey the same laws of addition , they do not obey the same laws of multiplication , for the product of two lators is always zero .
] Thus every motor consists of a rotor part and a parallel lator .
If Q is an octonion , MiQ stands for the rotor of the motor of Q and mQ for a coaxial rotor of the same magnitude and sense as the lator of the motor of Q. If MAB = 0 , either A and B are coaxial , or one of them is a lator parallel to the axis of the other , or they are both lators .
If MtAB = 0 , either A and B are parallel , or one of them at least is a lator .
If mAB = 0 , either ( p + p ' ) sin 6 + d cos 0 = 0 ( where d and 0 are as before , and p and p ' are the pitches of A and B ) , or one is a lator parallel to the axis of the other , or they are both lators .
If SAB = 0 , either they intersect perpendicularly , or one is a lator perpendicular to the axis of the other , or they are both lators .
If SiAB = 0 , either they are perpendicular , or one of them at least is a lator .
If sAB = 0 either ( p + p ' ) cos 0 = d sin 0 , or one is a lator perpendicular to the axis of the other , or they are both lators .
The necessary and sufficient condition to ensure that SABC = 0 is n 2 172 Mr. A. McAulay .
[ Dec. 12 , either two independent motors of the complex A , B , C are lators or XA + YB + ZC = 0 , where X , Y , Z are scalar octonions whose ordinary scalars ( S , X , \amp ; c. ) are not all zero .
If B and 0 have definite not parallel axes , XB + YC is any motor that intersects the shortest distance of B and C perpendicularly , where X and Y are arbitrary scalar octonions .
The analogue in octonions of the linear vector function of a vector in quaternions is called a commutative linear motor function of a motor .
It is not the most general form of a linear motor function of a motor .
The latter is called a general function .
If a commutative function is such that , acting on an arbitrary rotor through a definite point , it reduces the operand to a rotor through the same point it is called a pencil function , and the point is called the centre of the pencil function .
The geometrical relations between a pencil function and rotors through its centre are precisely the same as the geometrical relations between a linear vector function of a vector and vectors .
When a commiltative function degenerates into a lator function ( for all values of the motor operand ) the geometrical relations between the function and rotors in general are also precisely the same as the corresponding quaternion relations .
A general function involves thirty-six ordinary scalars , a general self-conjugate twenty-one , a commutative function eighteen , a commutative self-conjugate twelve , a pencil function twelve , and a self-conjugate pencil function nine .
[ In the case of the pencil function three of the scalars go to specify the centre . !
-ar , a general self-conjugate , is called an energy function when sEarE is not positive for any motor E ; it is -a partial or complete energy function , according as sE-srE is zero for some values of E or for none .
We here pass over for the most part those properties of the- commutative function which are immediately deducible from the fact that octonions are formal quaternions .
Let 0 for the present stand for a commutative function .
0 satisfies a cubic with scalar octonion coefficients .
This cubic generally has three definite scalar octonion roots ; but sometimes it has an infinite number , sometimes only one , and sometimes none at all .
The cubic ( called the 0X cubic ) , whose coefficients are the ordinary scalar parts of the coefficients of the 0 cubic , has an important bearing on the geometrical properties of 0 .
0 can always be put in a trinomial form analogous to the corresponding quaternion trinomial form .
If X is a root of the 0 cubic corresponding to a single root of the 0 !
cubic , 0\#151 ; X can be put in a binomial form , and there is a line such that if E be any motor coaxial with the line 0E = XE , so that 0E is coaxial with E. If X corresponds to a repeated root of the 0i cubic , this is not always true .
Octonions .
1895 .
] If X is root of the 0 cubic that cubic can always be put in the form ( 0-X)(02-N'0 + N ) = 0 , even when there is no second root .
If X corresponds to a single root of the 0 , cubic ( 02\#151 ; N'0 + N)E = 0 , if E is any motor which intersects a certain line perpendicularly .
If X corresponds to a repeated root of the 0 !
cubic , the statement is not always true .
Except when all the roots of the 0 !
cubic are equal there is always some line such that if E be any motor coaxial with the line , 0E is also coaxial with the line .
[ When all the roots of the 0 !
cubic are equal the statement is sometimes true and sometimes untrue .
] Let , now , 0 stand for a real commutative self-conjugate .
It can then always be put in the form\#151 ; 0E = -XiSEi-X'jSEj-X"ft , where X , X ' , X " are three real scalar octonions , and i , j , ft are three mutually perpendicular unit intersecting rotors .
The cubic is ( 0\#151 ; X)(0\#151 ; X')(0\#151 ; X " ) = 0 , so that it always has three real roots .
These may be taken as X , X ' , X " , even when it has an infinite number of roots .
In this last case the X , X ' , X " of the equation 0E = \#151 ; Xt'SEf\#151 ; ... .
have definite values which are called the principal roots of the cubic .
Thus , there are always three mutually perpendicular intersecting lines such that if E be a motor coaxial with any one of them , 0E is coaxial with E. If two , but not three , of the roots of the 0i cubic are equal , and the two corresponding principal roots of the 0 cubic are unequal , 0 cannot be put in the form\#151 ; 0E = MAEB + YE , where A and B are constant motors and Y a constant scalar octo-nion .
In all other cases 0 can be put in this form , and A , B , and Y. are real .
The cubic is { ( 0\#151 ; Y)2\#151 ; A2B2}{(0\#151 ; Y ) + SAB } =0 .
When neither A nor B is a lator and they are not parallel , the principal axes ( the three lines mentioned just now ) , of 0 are the shortest distance of A and B , and the two lines which bisect this shortest distance and also bisect the angles between A and B. A scalar octonion may be of any one of five types ; one type is zero and the other four may be called positive and negative , scalar octonions and positive and negative scalar convertors .
If A and B are two motors ( not parallel and neither a mere lator ) , such that SA0B , = 0 , A and B are said to be fully conjugate with regard to 0 .
If A , B , C are three such motors of which each pair is fully conjugate the number of the scalar octonions SA0A , SB0B , SC0C , of any type is the same as the number of the principal roots of the same 174 Mr. A. McAulay .
[ Dec. 12 , type .
Making this fact the basis of a classification , it follows that there are ( rejecting the case 0 = 0 ) , thirty-four different types of 0 .
If Y ' ' is a commutative self-con jugate , such that SiE0E is negative , and not zero for all motor values ( except lators ) of E , there are always three ( and generally only three ) real lines which are not all parallel to one plane such that any three real motors with these lines for axes form a fully conjugate set both with regard to 0 and 0 .
A commutative self-conjugate is never a complete energy function , and can only be a partial energy function when it degenerates into a lator function .
The properties of the general function are examined by the help of " Grassmann 's Ausdehungslehre .
" Motors are in Grassmann 's own geometrical interpretations quantities of the second order .
" Inner Multiplication , " and the thory of " Normals , " have in this case interpretations which depend on an arbitrarily chosen origin .
Hence Grassmann does not apply his theories concerning these two to motors ; but motors may be treated as quantities of the first order , and if they are , and if Sir Robert Ball 's meaning of " reciprocal " is identified with Grassmann 's meaning of " normal " a real motor with negative pitch is a " simple imaginary " quantity of the first order , and the " numerical value " of a lator or a rotor is zero .
Grassmann generally assumes his quantities to be real and always assumes that a quantity is zero when its numerical value is zero .
Hence nearly all his theorems require modification in our case but his methods are generally with some extensions applicable .
" Combinatorial Multiplication " receives several applications .
A particular scalar combinatornal product of six motors is much used and also a motor combinatornal product of five motors .
The latter is the motor ( with a certain definite tensor ) reciprocal to the five motors of the product .
By aid of these two products , the following results ( the first of which is contained in Ball 's " Screws " ) are established .
To every complex of order n there is a complex of order 6\#151 ; n reciprocal and no motor not belonging to the latter , is reciprocal to the former .
If ( w ) be a complex of order n , and ( 6 n ) an independent complex of order 6\#151 ; n , then if ( 6\#151 ; n ) is the complex reciprocal to ( n ) and ( n ) the complex reciprocal to ( 6\#151 ; n ) , ( n ) and ( 6 ) are independent complexes .
What is called " combinatorial variation " is a species of variation of which Grassmann 's " linear " and " circular " variations are special cases .
Particular cases of combinatorial variation are these two , and also " hyperbolic " and " conjugate " variations , the last being with reference to a given general self-conjugate function .
Let w be a given real general self-conjugate function .
The motors E of a given complex ( n ) of order n , for which sEsrE = 0 from what Sir Robert Ball calls a complex of the ( n\#151 ; l ) , the order and second Octonions .
1805 .
] degree .
Two motors , E and F , are said to be conjugate with reference to w , when sEsrF =,0 .
In the language of " Screws " they are conjugate motors of the complex of the second degree .
^E is what Ball calls the polar of E. When w = 1 conjugacy reduces to reciprocity , and every theorem relating to the former becomes one concerning reciprocal motors .
A real motor , E , is called a positive , negative , or zero norm , according as sE-srE is \#151 ; 1 , +1 , or zero .
When \#187 ; =la positive norm has positive pitch , a negative norm negative pitch , and a zero norm zero or infinite pitch , i.e , it is a rotor or lator .
( n ) can always be expressed in an infinite number of ways as a complex of n real independent conjugate norms .
The number of norms of any type ( positive , negative , or zero ) is definite , and the complex of zero norms is a definite complex .
If ( n ) be a given complex of order and ( m ) a given complex of order to included in ( n ) , another complex of order can always be found with the following properties , ( to ) and ( n\#151 ; m ) together make up ( n ) , so that ( n\#151 ; m ) is also included in ( n ) , and ( to ) and ( n\#151 ; m ) are independent , ( to ) consists of the positive norms Ai A2 ... .
, the negative norms Bj B2 ... .
, and the zero norms Cb C2 , lb D2 ... ... ... ... ( n\#151 ; m ) consists of the positive norms A/ A2 ' ... .
, the negative norms B/ B2 ' , ------ and the zero norms C/ C2 ' ... .
D/ Da ' ... ... The number of the norms Dx D2 ... .
is the same as the number of the norms D/ D2 ' ... ... ... .
All pairs of these norms are conjugate except the following pairs of zero norms ( Di D/ ) , ( Do D./ ) ... ... These last are such that sD^D/ = sDotrDo ' \#151 ; .
, .
\#151 ; \#151 ; l. The pair of motors Ai " , Bt " , deduced from any such pair of exceptional zero norms Dl5 D/ by the equations A/ ' = ( Dx + D/ )/ y2 , B/ ' = ( Di\#151 ; D/ )/ v/ 2 , are conjugate , and are respectively a positive and negative norm of ( n ) .
Thus ( n ) consists of the following set of conjugate norms : ( 1 ) positive , At ... .
, A/ ... .
, A/ ' ... .
, ( 2 ) negative , B^ ... , B/ . . . .
, B , " ... .
, ( 3 ) zero , Ci ... .
, C/ ... ( n ) and ( n\#151 ; to ) are called semiconjugate complexes .
Denoting the complex of Ax A2 ... .
by ( A ) , that of A. A2 ... .
Bi B2 ... .
by ( AB ) , \amp ; c. , it is proved that ( A'B'CC'D ) contains all the motoi'S of ( w ) which are conjugate to ( to ) or ( ABCD ) , and no others .
( A'B'CC'D ) is therefore called the conjugate of ( to ) .
The sum of the orders of ( to ) and its conjugate exceeds n by the order of ( C ) , which is the complex common to ( to ) and the ( definite ) zero norm complex ( CC ' ) of ( n ) .
The conjugate of the conjugate of ( to ) is not in general ( to ) itself .
176 Mr. A. McAulay .
[ Dec. 12 , When ( \#171 ; ) and ( m ) and nothing else is given the statements of the following table are established .
The following complexes are determinate .
The following complexes are arbitrary to the extent mentioned .
( C ) ( coo ( CCD ) ( CD ) ( ABCD ) ( ABCCD ) ( A'B'CCD ) ... ... . .
( ABA'B'CCD ) ... .
( C ' ) is any complex which with ( C ) makes up ( CC ' ) ( D ) " " ( C ) " ( CD ) ( AB ) " " ( CD ) " ( ABCD ) ( A'B ' ) \#187 ; " ( CCD ) " ( AD'CCD ) Also ( D ' ) is indeterminate to the extent implied by the statement:\#151 ; ( DD ' ) is any complex which contains ( D ) , is conjugate to ( ABA'B ' ) , and with ( ABA'B'CC ' ) makes up ( n ) .
The individual motors Ci , C2 ... .
may be taken as any independent motors forming the determinate complex ( C ) , and the individual motors Di , D2 ... .
may be taken as any independent motors which with Cj , C2 ... .
form the determinate complex ( CD ) ; similarly for C/ , C2 ' ... .
; but not similarly for D/ , 13/ ... ... The table is much simplified , ( 1 ) when is an energy function and ( 2 ) when nr = I and n \#151 ; 6 .
If nr is any self-conjugate and E is one of a set of six conjugate motor 's , such that sE-nrE = 0 , then -arE = 0 .
If nr is an energy function ( partial ) and E is any motor , such that sE-srE = 0 , then nrE = 0 .
nr ( a general self-conjugate ) can always be expressed in an infinite number of ways in the form arE = \#151 ; 2/ FsEF , where F is one of six real constant independent motors , and / a corresponding ordinary scalar , which may be zero .
The six motors which are reciprocal to each set of five of the F 's form a conjugate set .
When none of the / 's are zero nr~x has a unique intelligible meaning .
The F 's form a conjugate set with regard to nr"1 .
When ( n ) and ( to ) are given the numbers of motors of the different types A , A \\amp ; c. , are all definite numbers characteristic of nr .
There are thus when n \#151 ; 6 , two methods of classifying nr .
The number of types of nr ( rejecting the case nr = 0 ) , when classified according as it affects motors in ( to ) only , is 2 , 5 , 9 , 14 , 20 , or 27 , when to has the value 1 , 2 , 3 , 4 , 5 , or 6 respectively .
The number of types of nr , when classified according as it affects motors in ( to ) and in a semiconjugate complex ( 6\#151 ; to ) , is 27 , 77 , 125 , 145 , 125 , 77 , or 27 , when to has the value 0 , 1 , 2 , 3 , 4 , 5 , or 6 respectively .
( n ) being as before , nr being a given general self-conjugate , and Y Octonions .
1895 .
] a given energy function , the following n real independent motors , forming a conjugate set with regard to can always be found , and in general uniquely ; ( 1 ) M ... .
ax ... .
, positive norms with regard to ur ; ( 2 ) Bj ... .
/ Si , negative norms with regard to nr(3 ) Cx ... .
, zero norms with regard to nr .
The numbers in the two groups / ij ... .
are the same .
The motors Ax ... .
, Bx ... .
, Cx ... .
, \#171 ; x + / 3x ... .
, ______ form a conjugate set ( not norms in general ) with regard to \\gt ; r , and of these on\#151 ; ... .
, are zero norms with regard both to nr and \\r .
When there are any a 's and / S 's there are not n real independent motors forming a common conjugate system of nr and \Jr .
If -0 is a complete energy function there are no s and / S 's .
Several particular cases of the general theorem are examined , those of especial importance being when nr \#151 ; 1 .
If nr and nr ' are two general self-conjugates such that sEurE = sEVE for every motor E of a given complex , then sE^-F = sE-nr'F where E and F are any two motors of the complex .
In particular , any motors of the complex which form a set conjugate with regard to nr are also conjugate with l'egard to nr ' .
If ( n ) is a given complex of order n and ( 6\#151 ; a given independent complex of order 6\#151 ; n , and if nr is given then nr ' can be determined uniquely , so that sEx^Ex = sE'-srEj , w , E2 = 0 where Ex is any motor of ( n ) and E2 any motor of ( 6 \#151 ; n ) .
Also , whatever motor value E have nr'E belongs to the complex reciprocal to ( 6\#151 ; If 0 is a general ( not necessarily self-conjugate ) function , such that it reduces every motor of ( n)to a motor of ( \#187 ; ) , and reduces every motor of ( 6\#151 ; n)to zero , 0 satisfies an 7i the .
If the roots of this equation are a repeated r times , b repeated s times , \amp ; c. , where a , ... .
are all different , there are certain definite independent complexes ( r ) , ( s ) , ____included in and making up ( ) , of orders ( r ) , cor- responding to these roots .
In ( r ) is included a complex ( p ) , ( not always definite ) of order p , where p is any positive integer not greater than r , such that ( 0\#151 ; a)M .
= 0 for every motor A of ( p ) .
If p be any positive integer and be a scalar different from both a and b , ( 0\#151 ; e)^E is not zero , and belongs to the complex consisting of ( r ) and ( s ) if E itself belongs to that complex .
A similar statement is true of the complex consisting of any number of the complexes ( r)\gt ; ( s ) , --If ( 0\#151 ; e)^E = 0 ( pany positive integer ) for every motor E of a complex of order qincluded in ( n ) , is a root of the tic repeated at least q times , and the complex is included in the complex corresponding to this root .
( 0\#151 ; b)s ( 0\#151 ; cY ... .
E where all the roots occur their full number of times , except a which is absent altogether , and where E is a motor of ( \#171 ; ) belongs to the complex ( r ) , and by giving a suitable value to E it may be made any motor of ( r ) .
Mr. A. McAulay .
[ Dec. 12 , If ( n ) is_the reciprocal of ( 6\#151 ; and ( 6 is the reciprocal of ( n ) , then ( n ) and ( 6\#151 ; n ) stand towards 0 ' , the conjugate of 0 , in exactly the same way as ( n ) and ( 6\#151 ; n ) stand towards 0 .
The ( fit n-tic is the same as the 0 w-tic , and the complex corresponding to the root a of the 0 ' n-tic is reciprocal to ( \#163 ; ) , ... .
and to ( 6\#151 ; n ) .
If 0 be self-conjugate ( 0 ' = 0 ) the complexes corresponding to the different roots are both conjugate , with regard to 0 , and reciprocal .
When not one of the roots of the 0 tic is zero 0-1 applied to motors of ( n ) , and supposed to reduce them to motors of ( n ) , has a unique intelligible meaning .
It is , like 0 , a general function .
The roots of its n-tic are the reciprocals of the roots of the 0 n-tic .
The complex corresponding to any root of the 0_1 n-tic is the same as the complex corresponding to the corresponding root of the 0 n-tic .
01-1 is the conjugate of 0-1 .
If 0 is real the .coefficients of the n-tic are real , but not necessarily the roots .
If a.'and b are two corresponding imaginary roots ( i.e. , ab and a+b are real ) , r = s. The complexes ( r ) and ( s ) are then imaginary , but the complex consisting of ( r ) and ( s ) is real .
In this real complex is included a real complex ( 2p ) of order 2p , where p is any positive integer not greater than r , such that { ( 0\#151 ; a)(0\#151 ; 6 ) } ?
reduces every motor of ( 2p ) to zero .
A real general self-conjugate ( octonion ) vr differs from a quaternion self-conjugate , and differs from a Grassmann self-conjugate in that ( 1 ) the roots of its sextic ( n \#151 ; 6 ) may be imaginary , and ( 2 ) there may be not more than one motor , A , in the complex corresponding to a repeated root a , for which ( w\#151 ; A = 0 , whether a be real or imaginary .
The roots of an energy function sextic are always real , and when for such a function a is not zero , every motor A of ( r ) is such that ( 0\#151 ; a)A = 0 .
But when there is not even in this case necessarily more than one motor A for which 0A = 0 .
For a complete energy function a is never zero .
If 0 degenerates into a commutative function the roots of its sextic are the roots of what was above called the 01 cubic each repeated twice .
In particular , if 0 degenerates into a selfrconjugate commutative function the roots of the sextio are all real , and consist of three pairs of equal roots .
The three corresponding complexes are three sets of coaxial motors , whose axes are three mutually perpendicular intersecting lines .
Except when what were called the principal roots of the 0 cubic are all ordinary scalars , there are not in this case six co-reciprocal motors , which also form a conjugate set .
Combinatorial variation is applied to prove some other facts , w being a general self-conjugate , and ( \#171 ; ) , as before , the sum of the reciprocals of the pitches of SFxirFi , SF2wF2 , ... .
where Fi , F2 ... .
Oct , onions .
1895 .
] are ^ conjugate sot of n independent motors of ( ji ) is constant .
In particular , putting = 1 we get the theorem proved in " Screws , " that the sum of the reciprocals of n independent co-reciprocal motors of O ) is constant .
If A , B , C are any three independent motors of a complex of the third order the pitch of SABC is constant , and if A , B are any two independent motors of a complex of the second order the pitch of MAB is constant .
Or adopting the rotation above the sum of d cot 0\#151 ; etan 0 , and the pitches of A , B , C is constant in the first case , and the sum of d cotd and the pitches of A and B is constant in the second case .
The most general forms of complexes of all orders , expressed whenever possible as consisting of reciprocal motors with axes along mutually perpendicular intersecting lines , are given ; as also the reciprocal complexes in a similar form .
The differentiation of octonion functions is considered .
It is shown that physical problems may be treated in a manner which , so far as appears from the present trials is in most cases practically identical with quaternion method .
A symbolic rotor y is defined , which has properties very similar to the vector y of quaternions .
Integration theorems coi'responding to the well known quaternion ones are given .
The octonion treatment of strain and of intensities and fluxes is given at some length .
The octonion formulae , though bearing a somewhat more extended meaning than the quatei-nion formulae , are surprisingly similar in form to the latter .
On the whole , from this part of the paper , it cannot be said that octonions prove more efficient than quaternions , though what would appear were the subject more dev el oped cannot at present be said .
The following are some of the applications made in the last division of the paper :\#151 ; A twist means a general displacement of a rigid body .
A twist can always be effected in an infinite number of ways by two rotations .
The following construction suffices to find any two such rotations .
Take any line 1 intersecting the axis of the twist perpendicularly .
Let 1 become 2 when it is subjected to half the given twist , lake any transversal 3 of 1 and 2 .
Then double the rotation that converts 1 into 3 , followed by double the rotation that converts 3 into 2 , will effect , the given twist .
A right-about-turn means a rotation through two right angles ; thus it is completely specified by its axis .
A twist can always be effected in an infinite number of ways by two right-about-turns .
The following suffices to obtain the axes in all cases .
The axis of the first right-about-turn is any line intersecting the axis of the twist perpendicularly .
The second axis is obtained from the first by giving to the latter half the given twist .
As is well known , two equal parallel and opposite rotations combine into a translation .
The translation is compounded of two translations , the Mr. A. McAulay .
Octonions .
[ Dec. 12 , one being equal and parallel to the perpendicular from the first axis on the second , and the other equal and parallel to the same perpendicular when it has been first rotated with the second rotation and then reversed .
To combine two twists , take two lines , 1 and 2 , such that half the first twist brings 1 into coincidence with the shortest distance between the axes , and half the second twist brings the shortest distance into coincidence with two .
Then the axis of the resultant twist is the shortest distance between 1 and 2 and the twist itself double the twist about this axis , which will bring 1 into coincidence with 2 .
The geometrical properties of the second order and third order complexes , as given by Sir Robert Ball in " Screws , " are established .
The octonion treatment of the motion of a single rigid body is considered at some length .
The treatment is very analogous in many parts to the quaterion treatment of the motion of a rigid body with one point fixed .
* The variation in time of the pitch and position of the velocity motor of a rigid body subject to no external forces is considered .
This leads to the consideration of a curve\#151 ; called the " normal cone curve " \#151 ; related to the polhode .
The polhode lies on a quadric cone whose vertex is the centre of the Poinsot ellipsoid .
The normal cone curve lies on the quadric cone whose vertex is the same point , and which is normal to the polhode cone .
Defining a polhode as the locus of points on a quadric at which the tangent planes touch a concentric sphere , it is shown that a real polhode on a real quadric is always a polhode on a second real coaxial quadric , and that the normal cone curve is a polhode on each of two other coaxial quadrics which are both real or both imaginary .
The original polhode is similar and similarly situated to the normal cone curve of the original normal cone curve when the latter is regarded as a polhode .
Several reciprocal relations between the two pairs of quadrics are established .
If either of the original polhode quadrics is an hyperboloid of two sheets , the normal cone curve quadrics are imaginary , though the normal cone curve itself is always real .
When the four quadrics are all real , not more than three of , them can be ellipsoids , though they may all be hyperboloids of one sheet .
A figure drawn to scale is given in which each pair consists of an ellipsoid and an hyperboloid of one sheet , and in which each polhode is identical with the normal cone curve of the other .
The two polhode quadrics coalesce if , and only if , one is a sphere .
The general solution given breaks down in certain limiting cases .
These cases are examined , and all prove to have simple geometrical properties .
The methods adopted in these-particular operations are to all intents and purposes quaternion methods , and they illustrate , what appears frequently throughout the 1895 .
] The Formation and Structure of Dental Enamel .
181 paper , that quaternion methods may be regarded practically as a particular case of octonion methods .
When the motion of the rigid body is always in tbc neighbourhood of an absolutely stable position we get the theory of " Screws .
" The more general theorems in connection with this are given in a somewhat different form from Sir Robert Ball 's .
Thus , for instance , we have a " generalised force motor , " which in a particular case becomes the " reduced wrench .
" It is pointed out that the latter is not always intelligible .
Some form of the former is always intelligible , as also is a particular form of it called the " virtual force motor .
" The corresponding " virtual impulse motor " is the impulse motor due to the external impulses and the reactions of the constraints , but a similar statement does not hold for the virtual force motor IV .
" On the Formation and Structure of Dental Enamel .
" By J. Leon Williams , D.D.S. , L.D.S. Communicated by Professor Schafer , F.R.S. Received December 4 , 1894 .
( Abstract .
) The special points in the formation and structure of enamel which 1 have attempted to elucidate in this paper may be summarised as follows\#151 ; 1st .
The existence of a very thin membrane , or a structure of membrane-like appearance , lying between the ameloblasts and the forming enamel , and also between these cells and those of the stratum intermedium .
I have also , in many specimens , seen a similar membrane covering the odontoblasts .
2nd .
The formation of enamel by deposit and not by cell calcification .
This deposit probably consists of two distinct cell products\#151 ; a granular plasm and spherules of calcoglobulin .
3rd .
The relation of the cells of the stratum intermedium to true secreting tissue ; this relation being especially marked in the enamel organs of the rat and mouse .
4th .
An intricate vascular network in the stratum intermedium .
I should also mention that I have seen a free distribution of blood vessels in the odontoblastic layer of cells in the mouse , rat , and calf , as well as in human embryos , thus conclusively proving that these cells are not calcified .
5th .
The fibrous character of enamel in many of the lower animals , and the change of these fibres into more or less regularly arranged granules in the monkey and in man .
6th .
That the varicosities of the enamel rods are not caused by acids ( although often rendered more clear to view by acid treatment )

