What is error variance? The term error variance doesn't currently have either a tag here, or a page on Wikipedia. 
p259 of Colman (2015) defines error variance as:

In statistics, the portion of the variance in a set of scores that is
  due to extraneous variables and measurement error.



*

*Does the 'variance' part of the term imply that error variance represents the expectation of the squared deviation of a random variable from its mean?

*Does the 'error' part of the term imply that the random variable in question must be a estimate/prediction of some outcome?

*Is the term error variance therefore inappropriate unless we're dealing with a situation involving a) the estimation/prediction of some outcome, and b) the quantification of error in terms of squared deviations?

*Is Colman's definition a (near-)perfect one? Can anyone volunteer an improvement over it?
Colman, A. M. (2015). A dictionary of psychology. Oxford University Press, USA.
 A: You are quite right. In the context of linear regression, or of any other model that can yield predictions on one variable (response) from values of other variables (predictors), we usually have a set of observations, that is, points where we observed the actual response and the predictors. Given a model, for each observation we can compute the predicted value (from model and predictors) and the actual value. The error is the difference between predicted and observed value.
Since we have a set of observations, we have a set of errors and therefore we can compute its variance. Furthermore, if observations are seen as a random variable, we can estimate its variance.
That is error variance.
Colman's definition is equivalent to this one, but from another point of view.
In our set of observations, we can compute variance of the response. If we have a model, we can explain part of the variance of the response from the variance of predictors. The part we can't explain is error variance - the same error variance explained above.
And to summarize: Please have a look at the page of Corman's book and notice that just before "error variance" there is the definition of "error variable". A short definition of error variance is that it is the variance of the error variable.
