Volume 3 of Herman Bavinck's "Reformed Dogmatics" addresses the issue of Limited Atonement. Bavinck deals with the matter at #404-08 (pp. 455-75 in the Baker Academic 2008 printing). Bavinck approaches the matter in a way that I found very helpful in light of certain Amyraldian commentators of late, in that he discusses many of the historical aspects of the issues relating to the extent of the atonement.
Bavinck quite correctly notes that "Intensively the work of Christ is of infinite value but also extensively it encompasses the whole world." Standing alone, such a statement could easily be misinterpreted - and the incautious reader is advised to be careful to finish reading the entire selection before reaching hasty conclusions about Bavinck's position. Bavinck goes on to explain what is intended by the terms employed, as well as what is not intended, rejecting Origen's super-universalism and seemingly adopting Augustine's particularism.
Bavinck quotes Augustine as stating "everyone who has been redeemed by the blood of Christ is a human; yet not everyone who is a human has been actually redeemed by the blood of Christ." "Not one person perishes of those for whom Christ died." The footnotes identify Epistle 102 of Augustine as the source, although I have not confirmed that this is the case with reference to the originals.
The work is clearly an academic work, and some of its most valuable contributions are the footnotes, which in some cases point one to source material, and in other cases index important related works, such as Bellarmine's Controversies, Turretin's Institutes, and van Mastricht's Theology.
Bavinck appears to err somewhat on the issue of Creation at the beginning of section 407 (p. 470), where he ascribes the creation of the world distinctly to the Father, rather than the Son. Nevertheless, generally Bavinck's discussion seems reasonable, and was enjoyable. For example, although Bavinck acknowledges Augustine's positive contributions to our understanding of theology, Bavinck is not afraid to identify an error in Augustine's thought (his view that the number of elect men corresponds numerically with the number of fallen angels).
I would commend the twenty pages or so of necessary reading to those interested in further study of the atonement. I'd particularly commend this section to those Amyraldians (or quasi-Amyraldians) that have been trying to make arguments from historical theology, as well as trying to formulate a system of their own.