When did the meaning of the term "metabelian" change? I just realised that the meaning of the term "metabelian", when applied to groups, or Lie algebras, seems to have changed over years. (These days, it means that $[[G,G],[G,G]]$ is trivial, while in the past it was occasionally used to indicate that $[[G,G],G]$ is trivial. The difference here is that between solvability and nilpotence, that is.) 
This wiki says "The concept and term metabelian group was introduced by Furtwangler in 1930.
The term metabelian was earlier used for groups of nilpotence class 2, but is no longer used in that sense." (I don't understand "earlier" here. Can that sentence be parsed uniquely? Earlier than Furtwangler introduced the term? Earlier than the wiki article was written?)
I know at least one reference from mid 1960s (a PhD thesis from the US) where that old-fashioned usage is present, and I'd like to understand the history better in this instance. Thanks for help.
EDIT: it appears that the situation with usage and its history may be even different for groups and Lie algebras; I only dealt with literature on metabelian Lie algebras, and from answers so far I gather that there may be difference.
 A: I have the impression that even nowadays the term "metabelian" can be confusing if you talk about.
Although I would also say, that it means $2$-step solvable now most of the time, it still can mean $2$-step nilpotent sometimes. In particular, "metabelian Lie algebras" can still mean $2$-step nilpotent Lie algebras, see the following (and other) articles:
M. A. Gauger: On the Classification of Metabelian Lie Algebras (1973). 
E. M. Luks: What is the typical nilpotent Lie algebra ? (1977). 
L. Y. Galitski and D. A. Timashev: On the Classification of Metabelian Lie Algebras (1999). 
This is not an answer in the historical sense, but only a comment with too long references (which should perhaps belong to meta, if abelian belongs here).
A: The 1967 book Varieties of Groups by Hanna Neumann has the following footnote (p. 21): The term metabelian will always mean solvable of length two in agreement with current English usage; note however that in much of the Russian literature the term is used in the sense of `nilpotent of class two'.
