Currently reading through Weinberg's QFT book (Vol. 1) [readable in parts here].

In his derivation of the causal Dirac field in Ch. 5, he chooses his gamma matrices as (5.4.17) \begin{align} \gamma^0&=-i\left[\begin{array}{cc}0&1\\1&0\end{array}\right]\\ \gamma^i&=-i\left[\begin{array}{cc}0&\sigma^i\\\sigma^i&0\end{array}\right] \end{align}

As far as I can see this is just the Weyl basis as I know it from other sources (Peskin & Schroeder, Schwartz) but with an extra factor $-i$ here because of Weinberg's choice of metric $\eta^{\mu\nu}$=diag(-1,1,1,1) in the anticommutation relations for the gamma matrices


He then goes on to define the matrix

$$\beta=i\gamma^0=\left[\begin{array}{cc}0&1\\ 1&0 \end{array} \right] \,(i)$$

(both on p218 and pxxv) and uses this form of $\beta$ explicitly to derive the zero momentum spinors (5.5.17),(5.5.18).

Then, however, on the next page in front of Eq. (5.5.26), he states $\beta=-i\gamma^0$ and uses that in (5.5.26) and what follows to construct the causal field and spin sums. The form these equations take does seem to depend on the sign chosen for $\beta$ in (i). If I translate his spin sums back to the (more familiar to me) notation of P&S ($\gamma^\mu \rightarrow -i\gamma^\mu,a^\mu b_\mu \rightarrow -a^\mu b_\mu, \beta \rightarrow \gamma^0$) Eqs. (5.5.37),(5.5.38) have the familiar form they take in P&S.

So I'm wondering where the inconsistency went, because using both $\beta=\pm i \gamma^0$ in the same derivation should lead to one...?

[On a sidenote, in that section in Eqs. (5.5.20)-(5.5.23) the factors $(2\pi)^3$ are also all over the place and not consistent]


2 Answers 2


Ok, tracked the 2005 paperback edition down and indeed those parts have been corrected.

In front of Eq. (5.5.26) he now states $\beta=i\gamma^0$ and Eq. (5.5.26) has now changed to [2005 Paperback]

$$D(L(p))\beta D^{-1}(L(p)) = i{L_\mu}^0(p)\gamma^0=-ip_\mu \gamma^\mu /m,$$ with the sign change in the last step because of his $\eta^{00}=-1$ convention.

The previous edition [199x hardcover] reads

$$D(L(p))\beta D^{-1}(L(p)) = -i{L_\mu}^0(p)\gamma^0=-ip_\mu \gamma^\mu /m,$$ with an extra mistake in the last step giving the same answer so all the derivations after that remain correct. So this mistake in the last step was the inconsistency I was looking for and hadn't spotted.


The corrected edition of 1996 does not have such inconsistencies with the sign. Take the + sign and be sure to check the errata before continuing.

  • $\begingroup$ Is this the 2005 paperback edition you are referring to? I can't find any references to other updated (?) editions. Don't have access to the 2005 edition currently. I googled and checked the publisher site for errata lists for the first edition before posting my question but couldn't track anything down. If there is one available a link is always appreciated. $\endgroup$
    – kuzine
    Commented Nov 3, 2014 at 11:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.