In Higgins' paper _Baer invariants and the Birkhoff-Witt theorem_ (J. Algebra **11** (1969) 469–482, doi:[10.1016/0021-8693(69)90086-6](https://doi.org/10.1016/0021-8693%2869%2990086-6)) the following definition is given:

A _Lie structure_ over the $R$-module $M$ is a $T(M)$-bimodule $A$ together with a bilinear function $M\otimes M\to A$ taking $x \otimes y \mapsto \langle x,y\rangle$, satisfying 

- $\langle x,x\rangle = 0$;

- $\langle x,y\rangle t(uv-vu)=(xy-yx)t\langle u,v\rangle$, for all $x,y,u,v \in M$ and $t \in T(M)$; and

- $(\langle x,y\rangle z-z\langle x,y\rangle)+(\langle y,z\rangle x-x\langle y,z\rangle)+(\langle z,x\rangle y-y\langle z,x\rangle) = 0$, for $x,y,z \in M$.


My question is, how does this generalize the case of a Lie algebra over a field? And what is the motivation behind the second condition? Why can't we simply define a Lie structure over a ring to be an alternating bilinear law which satisfies the Jacobi identity?