McLaverty's project in this article is to delimit the boundaries of the application of authorial intention in modern scholarly editing. The nut of his theory of authorial intention is best laid out on pages 128-9. "The editor's concern", he says, "is with what not how the text is" (128). What McLaverty means to tell us is that while the editor should be concerned with, in fact is morally bound to consider, authorial intentions, he should do so in closely proscribed ways. McLaverty's memorable refrain throughout this section insists that the editor's aim is to "restore the text as the author intended to constitute it" (128). Roughly the first half of the article lays the theoretical groundwork for this characterization, and the last half deals with implications of it.

McLaverty relies on Tanselle, especially his 'The Editorial Problem of Final Authorial Intention', for the outline of the article. Tanselle's article did not put an end to the discussion of intentions, he says, only because:

1. the nature of intention has remained unclear and Hirsch's position continues to be misunderstood,

2. the important connection between intentions and meanings has involved motives in meaning unnecessarily, and

3. editors have endorsed practices which conflict with the principles they recommend (122).

For the section on pages 128-9, McLaverty cites Tanselle who draws on the work of Michael Hancher to enumerate three kinds of authorial intention: programmatic intentions, the author's intention to create a text which has certain qualities (say accuracy, elegance, popularity); active intentions, the author's intention write something in particular, or "be understood as acting in some way"; and final intentions, the author's intention to cause something to happen (perhaps a social movement). Of the three, only active intentions—the author's intentions to constitute the text of certain language—are in McLaverty's view of interest to the editor (128).

We can see the source of McLaverty's admonition that the editor's interest must focus on "those cases where the author's characterization of what he is doing (or meaning) is likely to be valid" in his earlier discussion of the three ways we commonly understand the concept of intent. We use intent in three ways: we talk about intentional actions (unimportant for the purposes of this paper), expressions of our intentions (characterizations of our actions), and the intentions with which we do something. And we learn from philosophy, he says, that we are more willing to accept a man's characterization of his own actions, i.e. his expressions of his intentions, the smaller the degree of observation we need to characterize it ourselves (123-4). (Note here that he seems not to say what he means. He says that you should be more willing to accept my account of my actions, the more the meaning of my actions are apparent to you. What he probably means is that you should be more willing to accept my account of my actions the more readily your observations support my account). McLaverty seems to want to apply this law of "parsimony of observation" to editing.

This preference for parsimony shows up in McLaverty's recommendations that the interpreter should attend to what Hirsch (depending on Husserl) called the author's active intentions, or verbal meaning, "that aspect of a speaker's 'intention' which, under linguistic conventions, may be shared by others" (124). With this endorsement of verbal meaning, McLaverty attempts to draw editors back to the written text and locate himself theoretically close to Husserl and Hirsch and distant from critics such as Poulet and Nehamas who look for some subjectivity not apparent in a single text. In appealing to Husserl, McLaverty also emphasizes that professional allegiance the editor owes to the author's original meaning is not based on intention but rather it is an ethical obligation. In other words, the editor restores the author's original meaning not because it is right or best or ideal but because we are morally bound to reflect what people actually say. (Note that in one of the more provocative statements of the text, McLaverty cites Husserl as claiming that for a text to be "determinate" and hence a possible "object of knowledge", by which he means certainty and not speculation, the meaning must be the author's meaning. None of these characterizations—that a text can be determinate, that it can be an object of certain knowledge, and that the meaning must be the authors or even that we can recover the author's meaning—seem readily obvious.)

The danger in editing on the basis of something other than active intentions, say programmatic intent or final intentions, is that editor would be attempting to fulfill intentions that were beyond the control of the author. The author has no control over the popularity of the text or how elegant or moving others perceive it to be. McLaverty uses Hayford and Sealt's editing of Melville's Billy Budd as an example of the problem with editing based on programmatic intent. When Hayford and Sealt edited Billy Budd, they emended Melville's description of a ship's crew (from Melville's 'a midshipman and two petty officers' to their 'a midshipman and two sailors') in order to make it more accurate (129). McLaverty's complaint with Hayford and Sealt is that by going beyond verbal meaning ñ or active intention ñ to programmatic intention, they have actually corrected not the score but the work itself. And the work should be beyond the reach of editing.

To build the theoretical groundwork for his claim that editors correct the work by following programmatic intention, McLaverty uses definitions for key concepts—score and performance—with shades of meaning different from those we've seen used by other theorists. McLaverty regards the text not as an instance of the work but as a score of it. "Those works of art which are types may have scores from which 'performances' of the work can be realized." If we accept that literature can be performed in the mind the way music can be, then we can say that the editor's concern is with the score.

The score represents the work, it is not identical with it, and the editor is entitled to alter the score to bring it in line with the work. The editor must not, of course, alter the workÖ but if the author has made a mistake in the score, the editor may correct it. The reader is not interested in the author's skill in writing a score, so the editor may supplement it...At times, of course, the editor's task will be very complicated because he will have to reconstruct a correct score from many defective ones (127).

This adherence to active intentions is both freeing and binding to an editor. It frees an editor from recovering what an author wrote and allows him to recover what the author intended to write. McLaverty dismisses two accounts of how the author is bound. He dismisses James Thorpe's proposal based on his definition that all art intends to achieve something, namely an aesthetic response, on the grounds that the definition of art is too broad and on the grounds that editing based on the author's attempt to achieve something is recourse to programmatic intentions. He also rejects, with Tanselle, simple formulations such as Bowers' project, "the recovery of the initial purity of an author's text", on the grounds that they break down in practice (125). He appeals once again to Husserl to describe what an editor is bound to. By interpreting Husserl's understanding of intentionality as "mental awareness" (123), he arrives at the formulation that the editor is bound to the reading the author "was aware of; if we had been able to stop him and ask him what he had written." (127). Notice how this formulation shades the meaning of intentionality from the Husserl's original "the direction or application of the mind to an object" to McLaverty's "what the author had in mind regardless of what he wrote".

This formulation provides McLaverty with the justification for the editor to correct authorial errors but only certain kinds. Accuracy, for McLaverty, is not an active intention, but a programmatic one and should be ignored by the editor. For the editor, therefore, the description of something as an error is not the important one. "The question is not, did the author intend this error? but, did the author intend this to constitute the text" (129).

We come to know the verbal meaning of a text through external evidence—the author's context though not his motives—and not through internal evidence—the text itself. McLaverty dismisses Tanselle's qualification of Hirsch: the work itself provides the best evidence of the author's intended meaning. "If the work were known, and the author's verbal meaning grasped, there would be no problem at all; the work of interpretation would have been done." The text is not an aid to interpretation; it is the object of interpretation (136). The editor who is committed to the author's meaning and intentions is committed to the author as a historical being (135). Problems of interpretation are settled by referring to context. And we explicitly cannot know the meaning of a text based on the author's motives. McLaverty is interested in the nature of specific actions executed in the text. And intentions express the nature of those actions while motives express only the spirit of the man that may or may not be realized in a particular work. (124). Note that in a seeming contradiction, McLaverty criticizes Tanselle for endorsing Gottlieb's attempt to discover (and edit out?) the parts of Great Expectations that Dickens wrote in attempt to please a friend, while he supports discarding changes in a text that an author makes under duress (137). Both of which seem to be motivations of the author external to the text.

Finally, McLaverty addresses final intentions. The view held by Tanselle and others that the editor should choose the final version because there is some overall intention evolving in the work can command no assent. The work, he says, is not a langue or a segment of a langue but the author's parole. And each issue of the work should be considered a new parole (131). Each version the author publishes should be considered a separate utterance and embody a new intention. Earlier versions present earlier intentions and meanings. McLaverty acknowledges that there can be some overall intention, i.e., to flatter the king or become poet laureate, but it would not be a basis for editing. Only active intentions constitute bases for editing (132). The editor, then, for the same moral reasons that compel him to edit to the author's verbal meaning, may choose among versions but may not alter the work (130).