The Photographic History

How this collection was made

What was searched, how each piece was pinned to a page, and how the transcriptions were read twice.

What was searched

All ten volumes of The Photographic History of the Civil War (New York: Review of Reviews Co., 1911), as digitized by the Internet Archive — three thousand six hundred and twenty leaves. Every page of every volume was scored for the vocabulary that can only mean Charleston harbor: Sumter, Moultrie, Wagner, Morris Island, Castle Pinckney, Secessionville, the Stono, the Swamp Angel, the Hunley, Cummings Point and a dozen more. Every page the search raised was then read.

What counts as a Charleston piece

The test is the subject, not the mention. A piece is here if what it is about is Charleston, South Carolina, or its harbor, defenses, garrisons, waters or prisoners, between the secession crisis of 1860 and the evacuation of February 1865. That rule excludes a good deal that scores highly: the chapter on the blockade, which is about the blockade of the whole Confederate coast and names Charleston as one port among Wilmington, Mobile, Savannah and New Orleans; the chronological tables of engagements at the back of volumes I, II and III, where Charleston appears as single lines in a list of the whole war; and the general index in volume X, which carries more Charleston lines than any chapter in the set and not a sentence of Charleston writing. Everything read and rejected is listed with the reason on the collection page.

The four shapes

This is a picture book, and its writing does not come in articles. Four things are published here, and each is labelled. A chapter is a chapter of a volume, entire — there are two, volume III’s Charleston, the Uncaptured Port and volume V’s Attack and Defense at Charleston. A passage from a chapter is a continuous run of narrative whose subject is Charleston, cut where the narrative turns to something else. A photograph description is the writing under a single plate: in this set that is a considered piece of prose of two or three hundred words, credited by volume to a named writer, and not a caption. A poem is given with the editors’ headnote to it and with the photographs the volume printed alongside.

How each piece was pinned to a page

Miller’s printers set the page number at the foot of the page, inside a printed ornament, and the scanner reads about one in eight of them. That is enough, because the numbering runs straight through each volume: a leaf is its printed page plus a constant, and the constant only changes where a scan gained or lost a leaf. So the numbers that can be read are used as observations of the constant, and no page is numbered unless legible numbers stand on both sides of it and agree — a leaf cannot have been gained or lost between two readings that agree, because gaining one would have changed the second. Of the 60 pages a citation here names, 59 are bracketed that way, with a confirming number never more than 31 leaves off. The constant was checked independently against the ten tables of contents: where the contents put a chapter on a page, the leaf this index assigns to that page carries that chapter’s title.

The tables of contents are used for nothing else, because they cannot be trusted. Volume V’s contents call the Charleston chapter The Defenses of Charleston and the chapter’s own title page reads Attack and Defense at Charleston; volume III’s say Charleston, the Unconquered Port where the book says the Uncaptured Port; and volume VI sets its titles in one column and its page numbers in another, so the scanner returns them out of register and every chapter in that volume comes back against the wrong page. Both titles are given on each piece’s page where they differ.

The transcriptions, and why they are read twice

Every page published here has been read by two different programs — the Internet Archive’s own reading of the scan, and a second reading made here from the page image at 1600 pixels — and the two are lined up against each other line by line. The Archive is good on the body type and poor on everything set large, because the titles of the photographs are in wide-spaced capitals printed across the picture itself. One title in volume III came back from the Archive as INSIDb miti MULLTRiE LOOKING EASTWARD I’l l.^llil, 1 OUT JOIIXSU:, and from the second reading as INSIDE FORT MOULTRIE—LOOKING EASTWARD OUTSIDE FORT JOHNSON.

Where the two readings agree, that is the text. Where they disagree, they are compared word by word, and a word is taken from the second reading only where the first reading’s word is not a word at all, the second reading’s is, the two are recognizably the same word, and one word stands against one word. Nothing is inserted, nothing is dropped, and the word a line breaks on is never touched. 165 lines on 51 pages carry a word mended that way. Nothing has been corrected by hand and nothing has been emended: where both readings fail, the damage stands, and it is visible. Every piece links to its own page in the scanned volume so that any passage can be checked against the printing.

Where the scan is plainly wrong on a point that matters — a date, a name — it is set out under The scan on that piece’s page; 4 pieces carry such a note.

What is likely missing

Two things. Captions set sideways on the page, which the scanner reads a letter at a time down the column and returns as nothing: volume I’s panorama of Charleston harbor is one, and it is listed among the rejected. And pages that print two unrelated descriptions in facing columns, where the scanner reads straight across both and the words of one run into the middle of the other; one Charleston plate is lost that way and is also listed. Anyone finding a piece that belongs here is asked to say so.

Rights

The set was published in 1911 and is in the public domain in the United States. The page images are the Internet Archive’s and are linked, not copied. Timrod’s two poems carry, in the 1911 volume, a permission line from the B. F. Johnson Publishing Company of Richmond as publishers of the Memorial Edition; that line is reproduced with the poems as the volume printed it. Timrod died in 1867 and the poems themselves have long been in the public domain.