Johnson’s defense

Editorial notes

What was checked, where the volume contradicts itself, and how a book with no transcription anywhere was read at all.

This page is the register for the Johnson collection: what was checked, what the volume gets right, where it contradicts itself, and what could not be settled from the two scans this section is built on.

The rule for the section is the site's rule. Nothing in Johnson's text is corrected. Where his book is wrong, or where it disagrees with itself, the disagreement is set out beside the passage rather than tidied out of it — because a source that has been silently tidied is no longer a source.

Johnson's own corrections

One of the two copies read for this section carries a Preface to the Second Edition, dated Charleston, 4 December 1890, in which Johnson lists what he wanted changed after six months in print. The Harvard copy does not have it. These are his corrections, in his order:

What	His correction
An omission in the Calendar	April 6th, (1863.) The Confederate transport, 'Marion' sunk by torpedoes while placing them in Ashley River. It is restored to its place on the calendar page.
The Hunley and the Housatonic	That the finding of the fish torpedo boat alongside the wreck of her victim is not a certified fact. His own chapter X states the discovery flatly; the second edition takes it back.
The Isaac Smith	That the captured gunboat, renamed the Stono by the Confederates, was lost in attempting to run the blockade.
Stono River, 16 July 1863	That in the attack on the Pawnee and Marblehead there were engaged four rifle guns of the siege-train (Palmetto Guards), Captain B. C. Webb, besides the artillery named on page 97.
Page 168	That the names Putnam and Strong should be transposed.
Fort Pulaski	That the average range of all the Union batteries at the reduction of Fort Pulaski was 2,559 yards, though the actual breaching was done at 1,700.

The register

3Correction

3Conflict

4To verify

3Confirmed

Correction

The Harvard copy is not the second edition

Front matter; Google Books Nu9lQ3uVC2YC / Internet Archive defensecharlest00johngoog

What is saidBoth copies carry a title page dated 1890 and are catalogued as the same book. The volume is usually cited as the 1890 second edition.

What the record showsThey are two different issues, and the difference is in the front matter. The Harvard copy, accession-stamped October 1890, has no preface to the second edition and no portrait plate of the author; it runs to 499 scanned images. The other copy has both, and 519. Since the added preface is dated 4 December 1890, the Harvard copy must be the earlier issue of that year.

What to doCite pagination freely — the text is identically paginated in both, and every citation on this site has been checked in both. But do not cite Johnson's own corrections from the Harvard copy: they are not in it.

Correction

The fort was not protected by its own rubble

Chapter VII, note at page 134; restated at page 264

What is saidThe received line — that the more Fort Sumter was battered the stronger it became — is repeated in almost every popular account of the siege.

What the record showsJohnson attacks it twice, and he is the man who did the work. The wasting of material was enormous; nothing but new material brought in for repairs could replace it; the debris alone could never have sufficed, however skilfully disposed. Sand, gabions, timber and bagged sand came down from the city by boat, at night, under fire, and the fort's own parade was mined four feet deep for sand before that.

What to doRetire the phrase. If a caption needs a line about Sumter's survival, it belongs to the seventeen filled rooms of the gorge and the nightly boats.

Correction

Fort Sumter did not cripple the New Ironsides

Chapter II, note at page 54

What is saidMilitary Operations of General Beauregard states that Fort Sumter crippled the New Ironsides on 7 April 1863.

What the record showsJohnson says he has failed to find any authority for it, and gives the contrary evidence: the ship was hit often at long range without damage, and five months later took seventy hits off Sullivan's Island in nearly three hours and her captain reported no damage at all.

What to doDo not repeat the claim on Roman's authority. Johnson is correcting his own side.

Conflict

The rounds fired in the first period of the first bombardment

Calendar, 23 August 1863; and the footnote table at page 143

What is saidThe calendar closes the first period with a total of 5,009 rounds.

What the record showsThe footnote table in chapter VII gives the two periods and their sum: second period 1,819, total 6,878. That makes the first period 5,059, not 5,009. The daily entries in the calendar for 17–23 August add to something close to but not identical with either figure, and the two scans differ over the calendar's own digit — one reads 5009, the other 6009.

What to doRead the total as approximate to within fifty rounds and quote whichever figure you cite to its own page. Settling it needs the printed volume and a careful re-addition of the daily entries.

Conflict

The Federal loss in the first assault on Battery Wagner

Calendar, 11 July 1863; and chapter IV, page 95

What is saidThe calendar gives the Union loss on 11 July as 339.

What the record showsThe text gives Strong's report of 436 casualties and captures for the two days, 10 and 11 July, of which 330 were directly referred to the assault. 339 appears nowhere in the text.

What to doCite the text's 330 for the assault and 436 for the two days, and note the calendar's variant rather than repeating it as a separate figure.

Conflict

How the 319 casualties divide

General Summary, page 273; and the paragraph at page 265

What is saidThe summary table reads: Total number of casualties (52 killed, 267 wounded) 319.

What the record showsThe paragraph two pages earlier reaches 267 by a different route — by deducting from 319 the casualties of the magazine explosion of 11 December 1863, which chapter X gives as 11 killed and 41 wounded, or 52. Both statements produce 52 and 267, and they cannot both be describing the same thing: one splits the total into killed and wounded, the other splits it into enemy action and accident.

What to doDo not print 52 killed and 267 wounded on the authority of this book alone. The total, 319, is the figure Johnson stands behind.

To verify

How many days the operations lasted

Calendar, 17–18 February 1865; and chapter XIII, page 259

What is saidThe last entry of the calendar gives the number of days of continuous military operations against Charleston. One scan reads 587, the other 667.

What the record showsThe text spells the figure out at the close of chapter XIII: upward of five hundred and sixty-seven days. That is the reading printed on this site. It is worth adding that 567 days back from 18 February 1865 reaches the end of July 1863, which corresponds to nothing in particular, while the descent on Morris Island, 10 July 1863, is 589 days back.

What to doQuote 567 with the page, and flag it. What Johnson was counting from is not stated anywhere in the volume.

To verify

Two pages exist in only one of the scans

Pages 258–259, chapter XIII

What is saidThe reading copy is used for every quotation on this site, and every quotation is confirmed against the second copy.

What the record showsThe reading copy's scan skips the leaf carrying pages 258 and 259 — the last two pages of the evacuation of Fort Sumter, including the passage on what was left standing and the moment the boats pulled away. The other copy has them. So the quotations from those two pages have been read by one machine only.

What to doThe one quotation on this site that comes from those pages is marked, in its citation, as carried by a single copy. Anyone quoting it for print should read the page image.

To verify

Appendix F begins at roman fifty, not at page 1

Contents page, page 12

What is saidThe contents page appears to give Appendix F — the reports, correspondence and despatches — as beginning at page 1.

What the record showsThe appendix is paginated in roman throughout, and the figure is a roman l, fifty, which both scans read as an arabic 1. The appendix runs on to page clxxvi by the Library of Congress record.

What to doCite appendix pages in roman. A citation to Appendix F, p. 1 is a citation to nothing.

To verify

The Hunley alongside the Housatonic

Chapter X, page 200; and the preface to the second edition

What is saidChapter X states that after the war the fish boat was found at the bottom close to the wreck of the larger vessel.

What the record showsJohnson withdrew it. The preface to the second edition says the finding is not a certified fact — and that preface is in only one of the two copies, so a reader working from the Harvard copy would never know he had withdrawn it.

What to doNever cite Johnson for the position of the Hunley. Cite him instead as an early witness that the story was doubted in Charleston in 1890.

Confirmed

The Swamp Angel burst on its thirty-sixth round

Chapter VI, note at page 132

What is saidNothing in Gillmore's own 1865 report records the end of the Marsh Battery gun.

What the record showsJohnson prints the inscription cut into the pedestal at Trenton, New Jersey, where the gun was set up after the war, which gives the weight of gun and projectile, the charge, the elevation, the opening date of 21 August 1863 — and burst 36th round.

What to doThis fills a gap in the Gillmore section, where the burst of the Swamp Angel is missing from the packet entirely. Cite Johnson, page 132.

Confirmed

The obstruction was not between Sumter and Moultrie

Chapter II, page 60

What is saidThe standard accounts place the harbor obstruction on a line between Fort Sumter and Fort Moultrie.

What the record showsJohnson says flatly that this is a mistake: it was placed between Sumter and Battery Bee, had dragged a little below the line of Battery Bee, consisted of parts of a broken boom and sections of rope-netting, carried no torpedoes at all on 7 April 1863, and left a gap three hundred yards wide next to Fort Sumter entirely clear.

What to doUse Johnson. He was in the fort, and the plates after his appendix show the layout.

Confirmed

The garrison of Fort Sumter on 7 April 1863 was 550

Chapter I, page 38; chapter II, page 44; calendar, 7 April

What is saidOne of the two scans reads the calendar's figure as 650.

What the record showsThe text gives seven companies of the 1st South Carolina Artillery, about 550 officers and men, at page 38, and 550 again at page 44 and in the summary at page 272. The other scan reads 550 in the calendar as well.

What to do550. The 650 is a misreading by a machine, not a variant in the book.

Method: two copies, and why

There is no transcription of this book. Project Gutenberg does not have it; Wikisource does not have it; the Internet Archive holds scans and machine-read text, and machine reading of 1890 type is wrong in a way that matters — not garbled beyond use, but wrong at the level of a single word, which is exactly the level a quotation lives at.

So two scans of two different copies are carried, and they are given different jobs. The copy with the cleaner reading of running prose is the one every quotation on this site was transcribed from, and it has to give the quotation word for word: anything less means the transcription is wrong. The Harvard copy is the check on it, and it is checked word by word rather than letter by letter, because its scan breaks individual letters constantly. Every word of a quotation four letters or longer has to appear on the matching page of the second copy, and any word that does not is named in the record.

Both copies must also agree on the printed page. The page is not read off the running head above the passage — a chapter opening carries no head, a plate carries none at all, and the scanner misreads them — but counted off the nearest readable head on each side and confirmed in the other copy. A phrase Johnson uses twice is matched at every place it occurs, and the copies have to agree on at least one of them.

A survey of the whole corpus runs alongside it. Of the 262 printed pages of text, 207 are carried by both copies in a form that can be compared at all — the rest are the volume's own plates, which have no text and no running head — and of those 207, the two copies agree on 199 and disagree on eight: pp. 62, 190, 191, 205, 207, 236, 262 and 272. No passage quoted on this site comes from any of those eight pages. Nothing that fails these checks is published. The quotation from pages 258–259, which only one copy carries, is marked as such where it appears. The checking runs at build time, from `_ingest/johnson/work/quotes.py`, and a quotation that stops matching stops the build.