DGR was able to carry out his alterations because of the arrangement he had made with his printer (Strangeways & Walden). As he wrote to Jane Morris (30 August 1869), he intended “to have the type of [the] sheets kept up and pay a rent for it. I find from the printer that this would not be very expensive” (see Fredeman, Correspondence , 69. 143 ). Under such circumstances DGR gained an enormous freedom to experiment with the composition of individual poems, groups of poems, and (finally) the whole volume as a bibliographical, rather than an inscriptional, process. Tennyson followed similar procedures with much of his poetry, but no one before Rossetti used this kind of process so elaborately, or at such textual depths.

The actual proof process began on or shortly before 21 July 1869, when DGR gave “Ellis the printer a number of scrappy poems and sonnets to print that I may keep them by me in an available form and perhaps be induced to do more” (see letter to Jane Morris, 21 July 1869, Fredeman, Correspondence , 69.91 ). At this point Ellis must have begun printing the proof state that would eventually be called the Penkill Proofs, which were ready for DGR around 18 August (this is the first integral set of page proofs, corresponding to what Lewis calls Proof State 2 (see Lewis, The Trial Book Fallacy , 186 ). While these proof pages were being printed Ellis seems also to have printed off a set of galley proofs that contained (at least) and .

As DGR worked over his poems in July and August his intentions began to shift significantly toward the idea of publishing a book. This change of mind is clear in his letter to Jane Morris of 30 August (see Fredeman, Correspondence , 69. 143 ). He was held back from committing to publication because he did not have good texts of several important poems. These were the works he had buried in his wife's coffin in 1862. So as he continued to correct the poems he did have, and to add some new things, he also began to see if the coffined volume might be recovered so that he could make copies of the poems he wanted (see his letters to Vernon Lushington and Charles Augustus Howell of 7 and 16 August, Fredeman, Correspondence , 69. 115. 126 ). He had Charles Augustus Howell act as his agent in this endeavor and the volume was eventually recovered on 5 October.

The proof process from July to the beginning of August is thus a distinctive one. It begins with the Penkill Proofs and continues through the so-called A Proofs and the A2 Proofs, the one printed off about 13 September, the other about a week later. These Lewis designates Proof States 3 and 4 (see Lewis, The Trial Book Fallacy , 186 ). The title DGR put at the head of these proofs, Poems. (Privately Printed.), indicates his idea about the project as a whole. DGR would retain this heading until December, at which point his determination to publish his work was completely settled. It is a determination that became fixed once DGR had recovered the lost poems and had satisfied himself that they were what he would want the public to see.

The dates of the A and the A2 Proofs indicate their volatile status. DGR was completely engrossed in the process of correcting and revising and adding work to the Penkill Proofs, a fact confirmed by his letters of August and September, particularly his letters to WMR. The A and the A2 Proofs represent a transitional correction phase that would culminate in the First Trial Book which Lewis designates as Proof States 5-7 ( The Trial Book Fallacy , 186-187 ). All the proofs to that point—approximately 3 October—share an important common bibliographical feature. DGR told his printers in late July or early August that “the longer poems can be printed on both sides,—only each poem must be separate—not with another poem on the back of the page. The poems which occupy only a page each must not have another on the back” (see letter to Strangeways of 7 August, Fredeman, Correspondence , 69. 117 ). DGR wanted to use the blank versos for corrections and additions, as he told Jane Morris.

The recovery of his poems from his wife's grave on 5 October marks the beginning of the next stage of the proof process. DGR immediately began copying out the poems he wanted and correcting these new copies. When this process was finished to his satisfaction he had the new poems printed off in the so-called Exhumation Proofs. This is a set of page proofs (67 pages in all) that contains the following works: (pages [1]-21), (pages 23-37), (pages 39-41), (pages 43-44), (page 45), and (pages 47-67). The Exhumation Proofs were printed off at the end of October. They correspond to Lewis's Proof States 8-9 (see The Trial Book Fallacy , 187 ).

That DGR had these poems printed in pages rather than galleys is consistent with the entire process. From the start he tried to have the types maintained in as modular a state as possible, so that he could add new works and augment or revise others with a minimal disruption of the standing types.

With these new poems in hand DGR decided to separate the prose tale from the text of his proofs. It was present from the outset, and with good reason, for the work is intimately related to DGR's entire corpus, being in fact a kind of critical exegesis, in fictional form, of his aims as a writer and artist. But his desire was to preserve the strictly poetic integrity of these writings if he could. He also wanted a relatively large book, however, so the removal of the tale wasn't feasible until he had the six new poems, in particular the three long ones. At that point was separated from the regular proofs. DGR had a few copies printed off for private circulation to friends. This is Lewis's Proof State 10 (see The Trial Book Fallacy , 187 ).

The final proof stage begins with the incorporation of the Exhumation Proofs into the First Trial Book, that is, with the printing of the Second Trial Book. The Second Trial Book proper was printed around 25 November, but proofs were pulled during the several previous weeks. Lewis designates these Proof States 11-12 (see The Trial Book Fallacy , 187 ).

The singularity of the Second Trial Book lies in this: that here, for the first time, the poems are all continuously paginated and printed. The Second Trial Book, unlike all the previous proof states, actually looks like an integral book. In fact, the surviving documents show clearly that the Second Trial Book provided DGR with the occasion for some major new ideas about the organization of the book that was now being certainly planned. Among the most notable of these is the decision to separate into the two sections of sonnets and songs, respectively, and the numbering of the units in each section. Also here is the first effort to organize the into a coherent order.

The next integral proof state is the Proofs for the First Edition. These proofs were printed about 1 March 1870 from a complex series of revisions, corrections, and additions to the accumulated previous sets of proof materials. This set of alterations was made on a variety of proof documents over a three month period. One gains a good insight into what was happening from a surviving set of Mixed Proofs for the First Edition designated by Lewis Proof State 13 (see The Trial Book Fallacy , 187 ). These mixed proofs contain documents from other proof states. They corroborate a fact we can also see in some of these other documents, in particular the Author's Working Copy of the Second Trial Book: that DGR kept working on copies that came from many different proof states, that he did not simply discard earlier proofs when a new proof state had been printed off. DGR's correspondence with Swinburne from late October 1869 into March 1870 also supplies a useful map of the complex revisions that were being contemplated and executed in this final proof phase.

Lewis designates the materials of this final stage Proof States 13-16 ( The Trial Book Fallacy , 187-188 ). It is helpful to see that in March, after the proofs were printed, a second state of the proof had to be pulled because so many late changes were being undertaken. Indeed, the process of altering continued into April, just before the book appeared, as DGR kept calling for changes that necessitated the printing of revises and even the shifting of pages.