# History of Topics in Special Relativity/Lorentz transformation (Quaternions)

## Lorentz transformation via quaternions and hyperbolic numbers

### Quaternions

The Lorentz transformations can also be expressed in terms of w:biquaternions: A Minkowskian quaternion (or minquat) q having one real part and one purely imaginary part is multiplied by biquaternion a applied as pre- and postfactor. Using an overline to denote quaternion conjugation and * for complex conjugation, its general form (on the left) and the corresponding boost (on the right) are as follows:[1][2]

{\displaystyle \left.{\begin{matrix}-x_{0}^{\prime 2}+x_{1}^{\prime 2}+x_{2}^{\prime 2}+x_{3}^{\prime 2}=-x_{0}^{2}+x_{1}^{2}+x_{2}^{2}+x_{3}^{2}\\\hline q'=aq{\bar {a}}^{\ast }\\\hline {\begin{aligned}q&=ix_{0}+x_{1}e_{1}+x_{2}e_{2}+x_{3}e_{3}\\q'&=ix_{0}^{\prime }+x_{1}^{\prime }e_{1}+x_{2}^{\prime }e_{2}+x_{3}^{\prime }e_{3}\\a&=\cos \chi +i\sin \chi =e^{i\chi }\end{aligned}}\\\left(a{\bar {a}}=1,\ \chi ={\text{imaginary}}\right)\end{matrix}}\right|{\begin{matrix}\chi ={\frac {1}{2}}i\eta \\\downarrow \\{\begin{aligned}x_{0}^{\prime }&=x_{0}\cosh \eta -x_{1}\sinh \eta \\x_{1}^{\prime }&=-x_{0}\sinh \eta +x_{1}\cosh \eta \\x_{2}^{\prime }&=x_{2},\quad x_{3}^{\prime }=x_{3}\end{aligned}}\end{matrix}}}

(7a)

After the parameter of 3D and 4D rotations were already given by Euler (1770/71), the quaternion transformation describing 3D rotations was given by Hamilton (1844/45) and Cayley (1845) and the corresponding 4D transformation was given by Cayley (1854, 1855). Cox (1882/83) discussed the Lorentz interval in terms of Weierstrass coordinates ${\displaystyle x_{0}^{2}-x_{1}^{2}-x_{2}^{2}-x_{2}^{2}=1}$ in the course of adapting w:William Kingdon Clifford's biquaternions a+ωb to hyperbolic geometry by setting ${\displaystyle \omega ^{2}=-1}$ (alternatively, 1 gives elliptic and 0 parabolic geometry). Stephanos (1883) related the imaginary part of w:William Rowan Hamilton's biquaternions to the radius of spheres, and introduced a homography leaving invariant the equations of oriented spheres or oriented planes in terms of w:Lie sphere geometry. Buchheim (1884/85) discussed the Cayley absolute ${\displaystyle x_{0}^{2}-x_{1}^{2}-x_{2}^{2}-x_{2}^{2}=0}$ and adapted Clifford's biquaternions to hyperbolic geometry similar to Cox by using all three values of ${\displaystyle \omega ^{2}}$. Eventually, the modern Lorentz transformation using biquaternions with ${\displaystyle \omega ^{2}=-1}$ as in hyperbolic geometry was given by Klein (1908-11) and Noether (1910) as well as Conway (1911) and Silberstein (1911).

### Hyperbolic numbers

Often connected with quaternionic systems is the w:hyperbolic number ${\displaystyle \varepsilon ^{2}=1}$, which also allows to formulate the Lorentz transformations:[3][4]

{\displaystyle {\begin{aligned}w'&=we^{-\varepsilon \eta }\\&=w(\cosh(-\eta )+\varepsilon \sinh(-\eta ))\\\\w&=w'e^{\varepsilon \eta }\\&=w'(\cosh \eta +\varepsilon \sinh \eta )\end{aligned}}\rightarrow {\begin{aligned}w&=x_{1}+\varepsilon x_{0}\\w'&=x_{1}^{\prime }+\varepsilon x_{0}^{\prime }\end{aligned}}\rightarrow {\begin{aligned}x_{0}^{\prime }&=x_{0}\cosh \eta -x_{1}\sinh \eta \\x_{1}^{\prime }&=-x_{0}\sinh \eta +x_{1}\cosh \eta \\\\x_{0}&=x_{0}^{\prime }\cosh \eta +x_{1}^{\prime }\sinh \eta \\x_{1}&=x_{0}^{\prime }\sinh \eta +x_{1}^{\prime }\cosh \eta \end{aligned}}}

(7b)

After the trigonometric expression ${\displaystyle e^{ix}}$ (w:Euler's formula) was given by Euler (1748), and the hyperbolic analogue ${\displaystyle e^{\varepsilon \eta }}$ as well as hyperbolic numbers by Cockle (1848) in the framework of w:tessarines, it was shown by Cox (1882/83) that one can identify ${\displaystyle ww^{\prime -1}=e^{\varepsilon \eta }}$ with associative quaternion multiplication. Here, ${\displaystyle e^{\varepsilon \eta }}$ is the hyperbolic w:versor with ${\displaystyle \varepsilon ^{2}=1}$, while -1 denotes the elliptic or 0 denotes the parabolic counterpart (not to be confused with the expression ${\displaystyle \omega ^{2}}$ in Clifford's biquaternions also used by Cox, in which -1 is hyperbolic). The hyperbolic versor was also discussed by Macfarlane (1892, 1894, 1900) in terms of w:hyperbolic quaternions. The expression ${\displaystyle \varepsilon ^{2}=1}$ for hyperbolic motions (and -1 for elliptic, 0 for parabolic motions) also appear in "biquaternions" defined by Vahlen (1901/02, 1905).

### Clifford algebra

More extended forms of complex and (bi-)quaternionic systems in terms of w:Clifford algebra can also be used to express the Lorentz transformations. For instance, using a system a of Clifford numbers one can transform the following general quadratic form into itself, in which the individual values of ${\displaystyle i_{1}^{2},i_{2}^{2},\dots }$ can be set to +1 or -1 at will:[5][6]

${\displaystyle {\begin{matrix}i_{1}^{2}x_{1}^{\prime 2}+\cdots +i_{n}^{2}x_{n}^{\prime 2}=i_{1}^{2}x_{1}^{2}+\cdots +i_{n}^{2}x_{n}^{2}\\\hline (1)\ x'=axa^{-1}\\(2)\ x'={\frac {ax+b}{\varepsilon ^{2}bx+a}}\end{matrix}}}$

(7c)

The Lorentz interval follows if the sign of one ${\displaystyle i^{2}}$ differs from all others. The general definite form ${\displaystyle x_{1}^{2}+\cdots +x_{n}^{2}}$ as well as the general indefinite form ${\displaystyle x_{1}^{2}+\cdots +x_{p}^{2}-x_{p+1}^{2}-\cdots -x_{p+q}^{2}}$ and their invariance under transformation (1) was discussed by Lipschitz (1885/86), while hyperbolic motions were discussed by Vahlen (1901/02, 1905) by setting ${\displaystyle \varepsilon ^{2}=1}$ in transformation (2), while elliptic motions follow with -1 and parabolic motions with 0, all of which he also related to biquaternions.

## Historical notation

### Euler (1770/71) – Quaternion rotation parameter

Euler (1770/71) demonstrated the invariance of quadratic forms in terms of sum of squares under a linear substitution and its coefficients, now known as w:orthogonal transformation. The transformation in three dimensions was given as:[M 1]

{\displaystyle {\begin{matrix}X^{2}+Y^{2}+Z^{2}=x^{2}+y^{2}+z^{2}\\\hline {\begin{aligned}X&=Ax+By+Cz\\Y&=Dx+Ey+Fz\\Z&=Gx+Hy+Iz\end{aligned}}{\begin{matrix}\left|{\scriptstyle {\begin{aligned}1&=AA+DD+GG\\1&=BB+EE+HH\\1&=CC+FF+II\\0&=AB+DE+GH\\0&=AG+DF+GI\\0&=BC+EF+HI\end{aligned}}}\right.\end{matrix}}\end{matrix}}}

in which the eight coefficients were related by Euler to four arbitrary parameter p,q,r,s:[M 2]

{\displaystyle {\begin{matrix}{\begin{aligned}A&={\frac {pp+qq-rr-ss}{u}}&B&={\frac {2pq+2ps}{u}}&C&={\frac {2qs-2pr}{u}}\\D&={\frac {2qr-2ps}{u}}&E&={\frac {pp-qq+rr-ss}{u}}&F&={\frac {2pq+2rs}{u}}\\G&={\frac {2qs+2pr}{u}}&H&={\frac {2rs-2pq}{u}}&I&={\frac {pp-qq-rr+ss}{u}}\end{aligned}}\\(u=pp+qq+rr+ss)\end{matrix}}}

which where rediscovered by w:Olinde Rodrigues (1840) who related them to rotation angles.[M 3]

They are equivalent to the parameter of an (improper) quaternion rotation in three dimensions, equivalent to (improper) Euler–Rodrigues parameters.

The case in four dimensions using sixteen coefficients was given by Euler as:[M 4]

{\displaystyle {\begin{matrix}V^{2}+X^{2}+Y^{2}+Z^{2}=v^{2}+x^{2}+y^{2}+z^{2}\\\hline {\begin{aligned}V&=Av+Bx+Cy+Dz\\X&=Ev+Fx+Gy+Hz\\Y&=Iv+Kx+Ly+Mz\\Z&=Nv+Ox+Py+Qz\end{aligned}}{\begin{matrix}\left|{\scriptstyle {\begin{aligned}1&=AA+RR+II+NN&0&=AB+EF+IK+NO\\1&=BB+FF+KK+OO&0&=AC+EG+IL+NP\\1&=CC+GG+LL+PP&0&=AD+EH+IM+NQ\\1&=DD+HH+MM+QQ&0&=BC+FG+KL+OP\\0&=BD+FH+KM+OQ&0&=CD+FH+LM+PQ\end{aligned}}}\right.\end{matrix}}\end{matrix}}}

Extending the Euler-Rodrigues parameter, he gave the corresponding representation of sixteen coefficients using eight parameters a,b,c,d,p,q,r,s[M 5]

${\displaystyle {\begin{array}{c|c|c|c}+ap+bq+cr+ds&+aq-bp+cs-dr&+ar-bs-cp+dq&+as+br-cq+dp\\\hline +aq-bp-cs+dr&-ap-bq+cr+ds&-as-br-cq-dp&+ar-bs+cp-dq\\\hline +ar+bs-cp-dq&+as-br-cq+dp&-ap+bq-cr+ds&-aq+bp-cs-dr\\\hline +as-br+cq-dp&-ar-bs-cp-dq&+aq+bp+cs+dr&-ap+bq+cr-ds\end{array}}}$
They are equivalent to the parameter of an (improper) quaternion rotation in four dimensions.

### Hamilton (1844/45) – Quaternions

w:William Rowan Hamilton, in an abstract of a lecture held in November 1844 and published 1845/47, showed that spatial rotations can be formulated using his theory of w:quaternions by employing w:versors as pre- and postfactor, with α as unit vector and a as real angle:[M 6]

(1) ${\displaystyle \beta '=\left(\cos {\frac {a}{2}}+\alpha \sin {\frac {a}{2}}\right)\beta \left(\cos {\frac {a}{2}}-\alpha \sin {\frac {a}{2}}\right)}$

In a footnote added before printing, he showed that this is equivalent to Cayley's (1845) rotation formula by setting[M 7]

(2) ${\displaystyle {\begin{matrix}\alpha \tan {\frac {a}{2}}=-\gamma \ \Rightarrow \ \beta '=(1+\gamma )^{-1}\beta (1+\gamma )\\\beta =ix+jy=kz,\ \beta '=ix_{\prime }+jy_{\prime }+kz_{\prime },\ \gamma =i\lambda +j\mu +k\nu \end{matrix}}}$.

Hamilton acknowledged Cayley's independent discovery and priority for first printed (February 1845) publication, but noted that he himself communicated formula (1) already in October 1844 to w:Charles Graves.

Formulas (1) or (2) are role models for Lorentz boost (7a), by replacing versors and quaternions with hyperbolic versors and biquaternions.

### Cayley (1845) – Quaternions

In 1845, w:Arthur Cayley showed that the Euler-Rodrigues parameters in equation E:(Q3) representing rotations can be related to quaternion multiplication by pre- and postfactors (an equivalent rotation formula was also used by Hamilton (1844/45)):[M 8]

{\displaystyle {\begin{matrix}x_{\prime }^{2}+y_{\prime }^{2}+z_{\prime }^{2}=x^{2}+y^{2}+z^{2}\\\hline (1+\lambda i+\mu j+\nu k)^{-1}(ix+jy+kz)(1+\lambda i+\mu j+\nu k)=\\{\scriptstyle ={\frac {1}{1+\lambda ^{2}+\mu ^{2}+\nu ^{2}}}\left\{{\begin{matrix}i\left[x\left(\lambda ^{2}+\mu ^{2}-\nu ^{2}\right)+2y(\mu \nu +\lambda )+2z(\lambda \nu -\mu )\right]\\+j\left[2x\left(\lambda \mu -\nu \right)+y\left(1-\lambda ^{2}+\mu ^{2}-\nu ^{2}\right)+2z(\mu \nu +\lambda )\right]\\+k\left[2x\left(\lambda \nu +\mu \right)+2y\left(\mu \nu -\lambda \right)+z\left(1-\lambda ^{2}-\mu ^{2}+\nu ^{2}\right)\right]\end{matrix}}={\begin{aligned}i(\alpha x+\alpha 'y+\alpha ''z)\\+j(\beta x+\beta 'y+\beta ''z)\\+k(\gamma x+\gamma 'y+\gamma ''z)\end{aligned}}\right\}}\\\downarrow \\{\begin{aligned}x_{\prime }=&i(\alpha x+\alpha 'y+\alpha ''z)\\y_{\prime }=&j(\beta x+\beta 'y+\beta ''z)\\z_{\prime }=&k(\gamma x+\gamma 'y+\gamma ''z)\end{aligned}}\end{matrix}}}

and in 1848 he used the abbreviated form[M 9]

${\displaystyle \Pi _{1}=\Lambda \Pi \Lambda ^{-1}\quad \ \left[{\begin{matrix}\Lambda =1+\lambda i+\mu j+\nu k\\\Pi _{1}=ix_{1}+jy_{1}+kz_{1}\end{matrix}}\right]}$

In 1854 he showed how to transform the sum of four squares into itself:[M 10]

{\displaystyle {\begin{matrix}x_{2}^{2}+y_{2}^{2}+z_{2}^{2}+w_{2}^{2}=x^{2}+y^{2}+z^{2}+w^{2}\\\hline MM'\left(w_{2}+ix_{2}+jy_{2}+kz_{2}\right)=(d+ia+jb+kc)(w+ix+jy+kz)(d'+ia'+jb'+kc')\\\left({\begin{aligned}M^{2}&=a^{2}+b^{2}+c^{2}+d^{2}\\M^{\prime 2}&=a^{\prime 2}+b^{\prime 2}+c^{\prime 2}+d^{\prime 2}\end{aligned}}\right)\end{matrix}}}

or in 1855 (where he also gave credit to Euler (1770)):[M 11]

{\displaystyle {\begin{matrix}\mathrm {x} ^{2}+\mathrm {y} ^{2}+\mathrm {z} ^{2}+\mathrm {w} ^{2}=x^{2}+y^{2}+z^{2}+w^{2}\\\hline \left(\mathrm {x} i+\mathrm {y} j+\mathrm {z} k+\mathrm {w} \right)=-{\frac {1}{\sqrt {MM'}}}(\alpha i+\beta j+\gamma k+\delta )(xi+yj+zk+w)(\alpha 'i+\beta 'j+\gamma 'k+\delta ')\\\left({\begin{aligned}M&=\alpha ^{2}+\beta ^{2}+\gamma ^{2}+\delta ^{2}\\M'&=\alpha ^{\prime 2}+\beta ^{\prime 2}+\gamma ^{\prime 2}+\delta ^{\prime 2}\end{aligned}}\right)\end{matrix}}}
Cayley's quaternion transformation of the sum of four squares, abbreviated QqP, served as a role model for the representation of the Lorentz transformation by Noether (1910), Klein (1910), in which the scalar part is imaginary.

### Cockle (1848) - Tessarines

w:James Cockle (1848) introduced the w:tessarine algebra as follows:[M 12]

${\displaystyle {\begin{matrix}t=w+ix+jy+kz\\t'=w'+ix'+jy'+kz'\\\left[i^{2}=-j^{2}=k^{2}=-1\right]\end{matrix}}}$.

While ${\displaystyle i^{2}=-1}$ is the ordinary imaginary unit, the new unit ${\displaystyle j^{2}=+1}$ led him to formulate the following relation:[M 13]

${\displaystyle \varepsilon ^{jy}=1+jy+{\frac {y^{2}}{2}}+j{\frac {y^{3}}{2.3}}+{\frac {y^{4}}{2.3.4}}+\&c.,\ \left(j^{2}=1\right)}$.
The real tessarine ${\displaystyle w+jy}$ is a w:hyperbolic number, ${\displaystyle \varepsilon ^{jy}}$ a w:hyperbolic versor, and the multiplication ${\displaystyle \varepsilon ^{jy}\left(w+jy\right)}$ leads to Lorentz boost (7b).

### Cox (1882) – Quaternions

w:Homersham Cox (1882/83) described hyperbolic geometry in terms of an analogue to w:quaternions and w:Hermann Grassmann's w:exterior algebra. To that end, he used w:hyperbolic numbers (without mentioning Cockle (1848)) as a means to transfer point P to point Q in the hyperbolic plane, which he wrote in the form:[M 14]

${\displaystyle {\begin{matrix}QP^{-1}=\cosh \theta +\iota \sinh \theta \\QP^{-1}=e^{\iota \theta }\end{matrix}}\left(\iota ^{2}=1\right)}$

In (1882/83a) he showed the equivalence of PQ=-cosh(θ)+ι·sinh(θ) with "quaternion multiplication",[M 15] and in (1882/83b) he described QP−1=cosh(θ)+ι·sinh(θ) as being "associative quaternion multiplication".[M 16] He also showed that the position of point P in the hyperbolic plane may be determined by three quantities in terms of Weierstrass coordinates obeying the relation z2-x2-y2=1.[M 17]

Cox's associative quaternion multiplication using the hyperbolic versor is equivalent to the Lorentz boost (7b) by setting ${\displaystyle Q=x_{1}^{\prime }+\iota x_{0}^{\prime }}$ and ${\displaystyle P=x_{1}+\iota x_{0}}$.

Cox went on to develop an algebra for hyperbolic space analogous to Clifford's w:biquaternions. While Clifford (1873) used biquaternions of the form a+ωb in which ω2=0 denotes parabolic space and ω2=1 elliptic space, Cox discussed hyperbolic space using the imaginary quantity ${\displaystyle {\sqrt {-1}}}$ and therefore ω2=-1.[M 18] He also obtained relations of quaternion multiplication in terms of Weierstrass coordinates:[M 19]

{\displaystyle {\begin{matrix}w^{2}-x^{2}-y^{2}-z^{2}=1\\\hline {\begin{aligned}PO^{-1}&=\cosh \theta +(li+mj+nk)\sinh \theta \\&=w+xi+yj+zk\\\\OP^{-1}&=\cosh \theta -(li+mj+nk)\sinh \theta \\&=w-xi-yj-zk\end{aligned}}\end{matrix}}}

### Stephanos (1883) – Biquaternions

w:Cyparissos Stephanos (1883)[M 20] showed that Hamilton's biquaternion a0+a1ι1+a2ι2+a3ι3 can be interpreted as an oriented sphere in terms of Lie's sphere geometry (1871), having the vector a1ι1+a2ι2+a3ι3 as its center and the scalar ${\displaystyle a_{0}{\sqrt {-1}}}$ as its radius. Its norm ${\displaystyle a_{1}^{2}+a_{2}^{2}+a_{3}^{2}+a_{4}^{2}}$ is thus equal to the power of a point of the corresponding sphere. In particular, the norm of two quaternions N(Q1-Q2) (the corresponding spheres are in contact with N(Q1-Q2)=0) is equal to the tangential distance between two spheres. The general contact transformation between two spheres then can be given by a w:homography using 4 arbitrary quaternions A,B,C,D and two variable quaternions X,Y:[M 21][7][8]

${\displaystyle XAY+XB+CY+D=0}$ (or ${\displaystyle X=-{\frac {CY+D}{AY+B}}}$).

Stephanos pointed out that the special case A=0 denotes transformations of oriented planes (see Laguerre (1882)).

The Lorentz group SO(1,3) is a subgroup of the conformal group Con(1,3) in terms of Lie's transformations of orientied spheres in which the radius indicates the fourth coordinate. The Lorentz group is isomorphic to the group of Laguerre's transformation of oriented planes.

### Buchheim (1884–85) – Biquaternions

w:Arthur Buchheim (1884, published 1885) applied Clifford's biquaternions and their operator ω to different forms of geometries (Buchheim mentions Cox (1882) as well). He defined the scalar ω2=e2 which in the case -1 denotes hyperbolic space, 1 elliptic space, and 0 parabolic space. He derived the following relations consistent with the Cayley–Klein absolute:[M 22]

{\displaystyle {\begin{matrix}\delta ^{2}+e^{2}\left(\alpha ^{2}+\beta ^{2}+\gamma ^{2}\right)=0\\\hline P=\delta +\omega \rho =\delta +\omega (\alpha i+\beta j+\gamma k)\\P'=\delta '+\omega \rho '=\delta '+\omega (\alpha 'i+\beta 'j+\gamma 'k)\\{\begin{aligned}PKP'&=(\delta +\omega \rho )(\delta '+\omega \rho ')\\&=\delta \delta '-e^{2}\rho \rho '+\omega (\rho \delta '-\rho '\delta )\\PKP&=\delta ^{2}-e^{2}\rho ^{2}\\\cos(PP')&={\frac {\delta \delta '+e^{2}\left(\alpha \alpha '+\beta \beta '+\gamma \gamma '\right)}{\left\{\delta ^{2}+e^{2}\left(\alpha ^{2}+\beta ^{2}+\gamma ^{2}\right)\right\}^{\frac {1}{2}}\cdot \left\{\delta ^{\prime 2}+e^{2}\left(\alpha ^{\prime 2}+\beta ^{\prime 2}+\gamma ^{\prime 2}\right)\right\}^{\frac {1}{2}}}}\end{aligned}}\end{matrix}}}
By choosing e2=-1 for hyperbolic space, the Cayley absolute becomes the Lorentz interval.

### Lipschitz (1885/86) – Clifford algebra

w:Rudolf Lipschitz used an even subalgebra of w:Clifford algebra in order to formulate the orthogonal transformation ${\displaystyle \varLambda X=Y\varLambda _{1}}$ of a sum or squares ${\displaystyle x_{1}^{2}+x_{2}^{2}\dots +x_{n}^{2}=y_{1}^{2}+y_{2}^{2}+\dots +y_{n}^{2}}$ into itself, for which he used real variables and constants, thus Λ becomes a real quaternion for n=3.[M 23] He went further and discussed transformations in which both variables x,y... and constants ${\displaystyle \lambda _{0}\dots }$ are complex, thus Λ becomes a complex quaternion (i.e. biquaternion) for n=3.[M 24] The transformation system for both real and complex quantities has the form:[M 25]

{\displaystyle {\begin{matrix}x_{1}^{2}+x_{2}^{2}+\dots +x_{n}^{2}=y_{1}^{2}+y_{2}^{2}+\dots +y_{n}^{2}\\\hline \varLambda X=Y\varLambda _{1}\\\left({\begin{aligned}X&=x_{1}+k_{1}k_{2}x_{2}+\dots +k_{1}k_{n}x_{n},\\Y&=y_{1}+k_{1}k_{2}y_{2}+\dots +k_{1}k_{n}y_{n},\\\varLambda &=\lambda _{0}+k_{1}k_{2}\lambda _{12}+\dots +k_{2}k_{3}\lambda _{23}+\dots +k_{1}k_{2}k_{3}k_{4}\lambda _{1234}+\cdots ,\\\varLambda _{1}&=\lambda _{0}-k_{1}k_{2}\lambda _{12}+\dots +k_{2}k_{3}\lambda _{23}+\dots -k_{1}k_{2}k_{3}k_{4}\lambda _{1234}+\cdots ,\end{aligned}}\right)\end{matrix}}}

Lipschitz noticed that this corresponds to the transformations of quadratic forms given by E:Hermite (1854) and E:Cayley (1855). He then modified his equations to discuss the general indefinite quadratic form, by defining some variables and constants as real and some of them as purely imaginary:[M 26]

${\displaystyle {\begin{matrix}{\begin{matrix}x_{1}={\mathfrak {x}}_{1},\dots x_{m}={\mathfrak {x}}_{m},\quad x_{m+1}=-h{\mathfrak {x}}_{m+1},\quad x_{n}=-h{\mathfrak {x}}_{m};\\y_{1}={\mathfrak {y}}_{1},\dots y_{m}={\mathfrak {y}}_{m},\quad y_{m+1}=-h{\mathfrak {y}}_{m+1},\quad y_{n}=-h{\mathfrak {y}}_{m};\end{matrix}}\\\lambda _{0}=\varrho _{0}\\\lambda _{ab}=\varrho _{ab};\quad a\leqq m,\ b\leqq m\\\lambda _{ab}=h\varrho _{ab};\quad a\leqq m,\ b>m\\\lambda _{ab}=h\varrho _{ab};\quad a>m,\ bm,\ b>m\end{matrix}}\ \left(h={\sqrt {-1}}\right)}$

resulting into

{\displaystyle {\begin{matrix}{\mathfrak {x}}_{1}^{2}+\dots +{\mathfrak {x}}_{m}^{2}-{\mathfrak {x}}_{m+1}^{2}-\dots -{\mathfrak {x}}_{n}^{2}={\mathfrak {y}}_{1}^{2}+\dots +{\mathfrak {y}}_{m}^{2}-{\mathfrak {y}}_{m+1}^{2}-\dots -{\mathfrak {y}}_{n}^{2}\\\hline P^{(m+1,\dots n)}{\mathfrak {\bar {X}}}={\mathfrak {\bar {Y}}}P_{1}^{(m+1,\dots n)}\\\left({\begin{aligned}{\mathfrak {\bar {X}}}&={\mathfrak {x}}_{1}+k_{1}k_{2}{\mathfrak {x}}_{2}+\dots +k_{1}k_{m}{\mathfrak {x}}_{m}-k_{1}k_{m+1}h{\mathfrak {x}}_{m+1}-\dots -k_{1}k_{n}h{\mathfrak {x}}_{n}\\{\mathfrak {\bar {Y}}}&={\mathfrak {y}}_{1}+k_{1}k_{2}{\mathfrak {y}}_{2}+\dots +k_{1}k_{m}{\mathfrak {y}}_{m}-k_{1}k_{m+1}h{\mathfrak {y}}_{m+1}-\dots -k_{1}k_{n}h{\mathfrak {y}}_{n}\\P&=\varrho _{0}+k_{1}k_{2}\varrho _{12}+\dots +k_{1}k_{2}k_{3}k_{4}\varrho _{1234}+\cdots ,\end{aligned}}\right)\end{matrix}}}
By setting m=n-1 or n=m+1, the Lorentz interval ${\displaystyle {\mathfrak {x}}_{1}^{2}+\dots +{\mathfrak {x}}_{m}^{2}-{\mathfrak {x}}_{n}^{2}={\mathfrak {y}}_{1}^{2}+\dots +{\mathfrak {y}}_{m}^{2}-{\mathfrak {y}}_{n}^{2}}$ and the Lorentz transformation follows

### Vahlen (1901/02) – Clifford algebra and Möbius transformation

Modifying Lipschitz's (1885/86) variant of Clifford numbers, w:Theodor Vahlen (1901/02) formulated Möbius transformations (which he called vector transformations) and biquaternions in order to discuss motions in n-dimensional non-Euclidean space, leaving the following quadratic form invariant (where j2=1 represents hyperbolic motions, j2=-1 elliptic motions, j2=0 parabolic motions):[M 27]

{\displaystyle {\begin{matrix}x_{0}^{2}-i_{i}^{2}x_{1}^{2}-\dots -i_{p}^{2}x_{p}^{2}=j^{2}\quad \left(i_{\alpha }^{2}=1,\ 0,\ {\text{or}}-1\right)\\\hline ax=ya'\\{\begin{aligned}a&=a_{0}+a_{1}i_{1}+\dots +a_{12}i_{1}i_{2}\dots +a_{12\dots p}i_{1}i_{2}\dots i_{p}\\a'&=a_{0}-a_{1}i_{1}-\dots +a_{12}i_{1}i_{2}\dots +(-1)^{p}a_{12\dots p}i_{1}i_{2}\dots i_{p}\\x&=x_{0}+x_{1}i_{1}+\dots +x_{p}i_{p}\end{aligned}}\\\hline {\text{Vector transformation}}\\y={\frac {ax+b'}{j^{2}bx+a'}}\\\left[z={\frac {cy+d'}{j^{2}dy+c'}},\quad y={\frac {ax+b'}{j^{2}bx+a'}}\right]\rightarrow z={\frac {Ax+B'}{j^{2}Bx+A'}}\\\hline {\text{Biquaternion}}\\(c+d'j)(a+b'j)=A+B'j\quad \left(i_{\alpha }j+ji_{\alpha }=0\right)\end{matrix}}}
The group of hyperbolic motions or the Möbius group are isomorphic to the Lorentz group.

### Klein (1908-11), Noether (1910) – Biquaternions

w:Felix Klein (1908) described Cayley's (1854) 4D quaternion multiplications as "Drehstreckungen" (orthogonal substitutions in terms of rotations leaving invariant a quadratic form up to a factor), and pointed out that the modern principle of relativity as provided by Minkowski is essentially only the consequent application of such Drehstreckungen, even though he didn't provide details.[R 1]

In an appendix to Klein's and Sommerfeld's "Theory of the top" (1910), w:Fritz Noether showed how to formulate hyperbolic rotations using biquaternions with ${\displaystyle \omega ={\sqrt {-1}}}$, which he also related to the speed of light by setting ω2=-c2. He concluded that this is the principal ingredient for a rational representation of the group of Lorentz transformations equivalent to (7a):[R 2]

{\displaystyle {\begin{matrix}V={\frac {Q_{1}vQ_{2}}{T_{1}T_{2}}}\\\hline X^{2}+Y^{2}+Z^{2}+\omega ^{2}S^{2}=x^{2}+y^{2}+z^{2}+\omega ^{2}s^{2}\\\hline {\begin{aligned}V&=Xi+Yj+Zk+\omega S\\v&=xi+yj+zk+\omega s\\Q_{1}&=(+Ai+Bj+Ck+D)+\omega (A'i+B'j+C'k+D')\\Q_{2}&=(-Ai-Bj-Ck+D)+\omega (A'i+B'j+C'k-D')\\T_{1}T_{2}&=T_{1}^{2}=T_{2}^{2}=A^{2}+B^{2}+C^{2}+D^{2}+\omega ^{2}\left(A^{\prime 2}+B^{\prime 2}+C^{\prime 2}+D^{\prime 2}\right)\end{aligned}}\end{matrix}}}

Besides citing quaternion related standard works such as Cayley (1854), Noether referred to the entries in Klein's encyclopedia by w:Eduard Study (1899) and the French version by w:Élie Cartan (1908).[9] Cartan's version contains a description of Study's w:dual numbers, Clifford's biquaternions (including the choice ${\displaystyle \omega ={\sqrt {-1}}}$ for hyperbolic geometry), and Clifford algebra, with references to Stephanos (1883), Buchheim (1884/85), Vahlen (1901/02) and others.

Citing Noether, Klein himself published in August 1910 the following quaternion substitutions forming the group of Lorentz transformations:[R 3]

{\displaystyle {\begin{matrix}{\begin{aligned}&\left(i_{1}x'+i_{2}y'+i_{3}z'+ict'\right)\\&\quad -\left(i_{1}x_{0}+i_{2}y_{0}+i_{3}z_{0}+ict_{0}\right)\end{aligned}}={\frac {\left[{\begin{aligned}&\left(i_{1}(A+iA')+i_{2}(B+iB')+i_{3}(C+iC')+i_{4}(D+iD')\right)\\&\quad \cdot \left(i_{1}x+i_{2}y+i_{3}z+ict\right)\\&\quad \quad \cdot \left(i_{1}(A-iA')+i_{2}(B-iB')+i_{3}(C-iC')-(D-iD')\right)\end{aligned}}\right]}{\left(A^{\prime 2}+B^{\prime 2}+C^{\prime 2}+D^{\prime 2}\right)-\left(A^{2}+B^{2}+C^{2}+D^{2}\right)}}\\\hline {\text{where}}\\AA'+BB'+CC'+DD'=0\\A^{2}+B^{2}+C^{2}+D^{2}>A^{\prime 2}+B^{\prime 2}+C^{\prime 2}+D^{\prime 2}\end{matrix}}}

or in March 1911[R 4]

{\displaystyle {\begin{matrix}g'={\frac {pg\pi }{M}}\\\hline {\begin{aligned}g&={\sqrt {-1}}ct+ix+jy+kz\\g'&={\sqrt {-1}}ct'+ix'+jy'+kz'\\p&=(D+{\sqrt {-1}}D')+i(A+{\sqrt {-1}}A')+j(B+{\sqrt {-1}}B')+k(C+{\sqrt {-1}}C')\\\pi &=(D-{\sqrt {-1}}D')-i(A-{\sqrt {-1}}A')-j(B-{\sqrt {-1}}B')-k(C-{\sqrt {-1}}C')\\M&=\left(A^{2}+B^{2}+C^{2}+D^{2}\right)-\left(A^{\prime 2}+B^{\prime 2}+C^{\prime 2}+D^{\prime 2}\right)\\&AA'+BB'+CC'+DD'=0\\&A^{2}+B^{2}+C^{2}+D^{2}>A^{\prime 2}+B^{\prime 2}+C^{\prime 2}+D^{\prime 2}\end{aligned}}\end{matrix}}}

### Conway (1911), Silberstein (1911) – Biquaternions

w:Arthur W. Conway in February 1911 explicitly formulated quaternionic Lorentz transformations of various electromagnetic quantities in terms of velocity λ:[R 5]

{\displaystyle {\begin{matrix}{\begin{aligned}{\mathtt {D}}&=\mathbf {a} ^{-1}{\mathtt {D}}'\mathbf {a} ^{-1}\\{\mathtt {\sigma }}&=\mathbf {a} {\mathtt {\sigma }}'\mathbf {a} ^{-1}\end{aligned}}\\e=\mathbf {a} ^{-1}e'\mathbf {a} ^{-1}\\\hline a=\left(1-hc^{-1}\lambda \right)^{\frac {1}{2}}\left(1+c^{-2}\lambda ^{2}\right)^{-{\frac {1}{4}}}\end{matrix}}}

Also w:Ludwik Silberstein in November 1911[R 6] as well as in 1914,[10] formulated the Lorentz transformation in terms of velocity v:

{\displaystyle {\begin{matrix}q'=QqQ\\\hline {\begin{aligned}q&=\mathbf {r} +l=xi+yj+zk+\iota ct\\q&'=\mathbf {r} '+l'=x'i+y'j+z'k+\iota ct'\\Q&={\frac {1}{\sqrt {2}}}\left({\sqrt {1+\gamma }}+\mathrm {u} {\sqrt {1-\gamma }}\right)\\&=\cos \alpha +\mathrm {u} \sin \alpha =e^{\alpha \mathrm {u} }\\&\left\{\gamma =\left(1-v^{2}/c^{2}\right)^{-1/2},\ 2\alpha =\operatorname {arctg} \ \left(\iota {\frac {v}{c}}\right)\right\}\end{aligned}}\end{matrix}}}

Silberstein cites Cayley (1854, 1855) and Study's encyclopedia entry (in the extended French version of Cartan in 1908), as well as the appendix of Klein's and Sommerfeld's book.

## References

### Historical mathematical sources

1. Euler (1770/71), pp. 75f
2. Euler (1770/71), p. 101
3. Rodrigues (1840), p. 405
4. Euler (1770/71), pp. 89–91
5. Euler (1770/71), p. 102
6. Hamilton (1844/45), p. 13
7. Hamilton (1844/45), p. 14
8. Cayley (1845), p. 142
9. Cayley (1848), p. 196
10. Cayley (1854), p. 211
11. Cayley (1855b), p. 312
12. Cockle (1848), p. 437
13. Cockle (1848), p. 438
14. Cox (1882/83a), pp. 85–86
15. Cox (1882/83a), p. 88
16. Cox (1882/83b), p. 195
17. Cox (1882/83a), p. 97
18. On pp. 104-105 he started using the term v2=-1, on p. 106 he noted that one can simply use ${\displaystyle {\sqrt {-1}}}$ instead of v, and on p. 112 he adopted Clifford's notation by setting ω2=-1.
19. Cox (1882/83a), pp. 108-109
20. Stephanos (1883), p. 590ff
21. Stephanos (1883), p. 592
22. Buchheim (1885), p. 309
23. Lipschitz (1886), pp. 75–79
24. Lipschitz (1886), pp. 134–138
25. Lipschitz (1886), p. 76; p. 137
26. Lipschitz (1886), pp. 145–147
27. Vahlen (1902), pp. 586–587, 590; (1905), p. 282

### Historical relativity sources

1. Klein (1908), p. 165
2. Noether (1910), pp. 939–943
3. Klein (1910), p. 300
4. Klein (1911), pp. 602ff.
5. Conway (1911), p. 8
6. Silberstein (1911/12), p. 793
• Klein, F. (1911). Hellinger, E.. ed. Elementarmethematik vom höheren Standpunkte aus. Teil I (Second Edition). Vorlesung gehalten während des Wintersemesters 1907-08. Leipzig: Teubner.
• Silberstein, L. (1912) [1911], "Quaternionic form of relativity", The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 23 (137): 790–809, doi:10.1080/14786440508637276

### Secondary sources

1. Synge (1972), pp. 13, 19, 24
2. Girard (1984), pp. 28–29
3. Sobczyk (1995)