Shillingsburg's questions about authorial intention fall into two categories: one is the nature of recoverability of intention and the other is the ontological status of literary works. Therefore, he addresses these issues separately in the two chapters "Intention" and "Ontology." Within the first of these categories, Shillingsburg asks two questions: 1. Is intention recoverable? 2. Is intention one thing that a single edited text can be made to conform to? This chapter answers a qualified and surprising "yes" to the first question – intention is recoverable if one uses his definition of intention. Shillingsburg answers "no" to the second question – no one text can conform to intention as he defines it.

Shillingsburg begins with Tanselleís long standing concept on intention that while intention can never be reached with assurance, one can strive toward it. Shillingsburg rejects this since how, he asks, can we know we are moving toward intention if we cannot know the thing (intention) we are moving toward.

Shillingsburg, like many critics, agrees that intention is a term used to mean many things. Many of its meanings are unstable, and worse, they define something that cannot be recovered. One level of intention stands apart in Shillingsburgís mind: the "intention to do" which underlies his position on the recoverability of intention:

"An intention to record on paper or other medium, a specific sequence of words and punctuation according to an acceptable or feasible grammar or relevant linguistic convention is specific and singular." (33)

Shillingsburg distinguishes between "the intention to mean" and "the intention to do." He positions his concept of "intention to do" close to Hancherís definition of "active meaning," but he believes Hancherís model still leaves too much room for critical interpretation and therefore non-absolute recovery. For example, he also thinks Hancherís "active intention" oversimplifies the production of a work over time when an author may alter and refine his "active intention" many times.

The "intention to mean" is inconclusively recoverable through critical interpretation. It is the "intention to do" that is absolutely recoverable – with three qualifications. These qualifications are the first sign of weakness in his definition and undermine his position on the recoverability of intention. Three things, he says, stand in the way of absolute recoverability: "scribal errors, ëFreudian slips,í and shorthand elisions" (33). So while we can never be sure what effect the author wanted his message to have –he may have wanted the message to mislead or be sarcastic for example –we can recover his intention to place words and punctuation on paper. It is this gap between the idea and the performance that Greetham will pry apart into a chasm in Theories of the Texts. Greetham believes that Shillingsburg acknowledges the gap between idea and performance but uses performance –the words actually on the paper –to recover the idea.

Shillingsburg himself widens the crack in his own foundation when he adds another definition of intention –the intention intended or "the text the author wished his audience to read" (36). Shillingsburg acknowledges this added distinction for those cases where it is believed that the author expected his editor and/or print setter to make certain modifications to his work. He intended and expected alterations, so the "intention to do" is not indicative of the idea that the author had for his published document.

The end of this theorizing on the intention of a text is this editorial position: "Presenting information in an orderly form, not just establishing a single authenticated text, is the editorial function" (38). Shillingsburg concludes his chapter by stating that intentions are multiple in nature and multiple "intentions to do" are concrete enough (evidenced in physical, material texts and manuscripts and supporting documentation) to give the careful editor firm ground for his work. Editors should provide a chronological arrangement of variants distinguished sociologically, historically, and authorially.

In the next chapter, Shillingsburg talks about the ontology of a literary work. As it provides definitions of terms that run counter to other theorists and therefore can complicate comparisons between them, I include the high points:

1. A work has no substantial existence nor does a version.

2. Since intentions change as time passes, the new revised document is separate from the preceding one and each represents only one envisioned version (recall a version is not material), which in turn is only one historically positioned view of the whole ideal work. (Shillingsburg does not mean "ideal" in the Platonic sense, but in the sense of unknowable in its totality.)

3. The text is contained and stabilized by the physical form but is not physical form. Each text, for Shillingsburg, represents, more or less well, one version.

4. The physical form is called a document.