The Secret War for the Union: The Untold Story of Military Intelligence in the Civil War

by Edwin C. Fishel; Boston, Houghton Mifflin, 1996

This is a title with a big promise, perhaps bigger than even 734 pages can deliver on. Raising expectations even further is Stephen Sears' dramatic introduction:

In the summer of 1959, at the National Archives in Washington, Edwin C. Fishel made a discovery that was, in the context of Civil War historical research, sensational. In what he has described as a "half roomful" of miscellaneous records of the Army of the Potomac were bundles of documents he had no idea still existed -- the operational files of that army's Bureau of Military Information, arranged with bureaucratic thoroughness and neatly tied with red tapes [sic]. They had been undisturbed for nearly a century. Fishel had discovered the first building block for what would be -- and is -- the first authentic history of military intelligence in the Civil War.

Secret War coverThe dust jacket takes us to even higher levels of expectation:

At the end of the war most of the intelligence records disappeared, and they remained hidden for almost a century, until Edwin Fishel uncovered them during the forty years of research that has resulted in this monumental book. The Secret War for the Union is unique among Civil War histories in its reliance on original, previously unknown sources.

The author's comment on sources, although not among the first parts normally read, further excites us with claims to lost knowledge:

The [archived intelligence] files were raided in 1962 ... about 75 inventory items, some of which may be bulky ones, remain unrecovered. The present writer had copied many of those items before the theft; the ones not copied may be counted as lost to history. In more recent years researchers have used the collection so recklessly that it has become thoroughly disarranged, bearing little resemblance to the inventory made at the time of its discovery in 1959. [...] It is now impossible to ascertain whether there have been losses in addition to those sustained in the 1962 theft.

One can hardly wait to start reading. But when the reading begins, this is what one finds: (1) It is not about Civil War intelligence, but largely about Army of the Potomac intelligence only, with a little bit added from other Eastern intelligence projects; (2) It starts with Scott and McClellan and ends without explanation at Gettysburg (3) A little summary that covers the Grant/Meade period is tacked on in a short epilogue; (4) Much of the material presented is familiar and based on known sources.

Contrary to what the dust jacket says and what Sears implies, the Bureau files Fishel discovered in 1959 mainly cover Hooker's regime at the head of the AoP, with nearly nothing from the preceding periods and perhaps just a little bit afterwards. This may explain the peculiar scheme of Secret War. Fishel first uses a large number of known sources to build the story up to the Hooker period; he then unveils a new, extended appreciation of Hooker based on his special sources; he then ends the book with a summary view of Meade and Grant as legatees of Hooker's excellent system.

Hooker liked good living and high times, so let's share a joke in his memory. A man sees his drunken friend circling a lampost at night. "What are you doing there Bill?" "I'm looking for my house key." "But you lost it in the tavern, Bill." "I know, but there's more light here."

The author of Secret War is circling a lampost called "Files of the Bureau of Military Information." Or is he?

In his "Comment on Sources," the author says he is: "This book violates one of the canons of modern works on the Civil War: it is not drawn from an enormous list of sources. The bulk of the intelligence story comes from a very small number of documentary collections."

Let us look at how a typical Hooker chapter is put together. Chapter 14 is a random pick. Here is a breakdown of its citations.








*These are the materials alluded to by Sears and the dust jacket.  * Good old OR  * Includes Douglas Freeman, etc.  * Includes newspaper clips, Sears' volume of McClellan letters, etc.

The proportion of material from BMI files plus archives, etc., almost reaches "bulk" status described by the author. He is circling the lampost here. But Hooker chapters do not make the bulk of this book.

Chapter 8, chosen at random, is a non-Hooker chapter dealing with Gen. John Pope: In 95 end notes (some of which have multiple citations), 72 cite the good old Official Records and at least 26 cite secondary sources, among whom Douglas Freeman is named no less than 12 times and John Hennesy six times. This chapter, which attempts to revise some of the conventional thinking on Pope, cites special archives six times only. Crudely formulated, less than 5% of the identified and labeled content comes from Fishel's special sources. Bulk? He has left his lampost, which is a good thing for a scholar to do, except it voids his "bulk" claim, and Sears' introduction, and the publisher's promotional text.

What Fishel has really done is use his finds to spotlight Joe Hooker's operations, after which point mass-market flashlights are brought in (Freeman, Sears, Hennesy, etc.) to supplement the glow of the OR moon. This is simplifying a bit, but is the gist of the case.


Given its limits, in comparison with its promise, readers may nevertheless be excited by the hope that the many unanswered intelligence questions in the Eastern Theatre will be illuminated. Primary among these is the mystery of the innacuracy of George B. McClellan's intelligence estimates of enemy strength.

McClellan's first official use of an opposing strength number, after arriving in Washington, appeared in an August 8 letter to his boss, Winfield Scott. From the treatment of this item, we see what Fishel has in store for McClellan:

The first of the famous exaggerations of Confederate strength was the 100,000 figure McClellan gave in his August 8 letter ...

Thus, Fishel starts with a conclusion (exaggeration) and then works toward the evidence. Evidence would have to include two numbers: one number would be the actual ("true") Confederate strength on August 8 near Washington, the other number would be the raw or base figure that McClellan supposedly puffed up to 100,000. We get neither. We get this:

The spies that McClellan said were among the sources of his 100,000 estimates could have been members of the short-lived bureau headed by William C. Parsons...

Now that the conclusion is set, the guessing begins. Because Parsons was disbanding his group at that time, Fishel quickly (and carelessly) dismisses them as a possible source for McClellan's figure.

Thus it is more likely that the spies McClellan referred to were self-appointed amateurs.

This is Fishel's tautology: (A) Parsons is in the process of disbanding his group THUS (B) Self-appointed amateurs replaced them as sources. We never learn which Federal informants might have earned Fishel's colorful, highly specific label (self-appointed amateurs) nor do we learn if such were reporting to McClellan. Yet somehow Fishel "knows" that McClellan "likely" relied on "self-appointed amateurs" -- just the kind of source to reflect badly on his professional judgement. Not satisfied entirely with the amateurs, Fishel rolls out another guess:

However, the likeliest single source was not a spy but a Confederate deserter, a Kentuckian named Edward B. McMundy.

Fishel does not give a McMundy-sourced strength figure. He says McMundy met with McClellan, Scott, Lincoln, and Seward, and that McMundy said McClellan praised him with having saved Washington, although "some people thought my views of the resources of the rebels to be extravagant." And so Fishel, having garlanded McClellan with self-appointed amateurs, now fastens McMundy around McClellan's neck. The logic seems to be that McMundy's figures were "extravagant" and therefore the exaggerator would have liked them.

For a taste of reality, let us leave Fishel and return to the letter. McClellan actually says in this August 8 letter, "Information from various sources, reaching me today, through spies letters and telegrams confirm my impressions derived from previous advices, that the enemy intend attacking our positions on the other side of the river as well as to cross the Potomac north of us." (This by the way was the pet project of a certain T.J. Jackson at that time.) Fishel has mischaracterized the letter: McClellan did not cite "spies letters and telegrams" as the source of his 100,000 estimate, but as the source of his belief in an upcoming attack. That is a very bad mistake.

McClellan continues, "I have also today received a telegram from a reliable agent just from Knoxville Tenn. that large reinforcements are still passing through there to Richmond. I am induced to believe that the enemy has at least 100,000 men in our front." Strictly speaking, no source (not spies, not letters, not telegrams) is cited for the 100,000. However, it is not unreasonable to infer that the agent's telegram helped him to make the estimate of 100,000. If this interpretation is possible, then McClellan is identifying one source (a Knoxville agent), and we do not have to speculate about McMundy. If McMundy had previously met Scott, McClellan could refer to McMundy in the letter as another source but he does not do that.

Fishel has misinterpreted a letter, then substituted a guess for a given in order to sustain a worthless conclusion.


Here's a choice Lincoln anecdote, interpreted by Fishel:

And Lincoln certainly had McClellan in mind when he said that the South must have a million men in its army, because he had half a million men in his own army and each of his generals claimed to be opposed by twice his own number.

Lincoln had "each of his generals" in mind, not just McClellan. The estimation errors apply to all generals on all fronts, not just to the AoP. Fishel inserts "McClellan" into Lincoln's mind in order to get the desired outcome: a Lincoln joke about McClellan's own estimates. This forcing represents a vice that has other manifestations, for instance rigging phony confessional testimony from a dead subject.


The ardent Republican general (and later historian) Jacob Cox long ago speculated that the estimates produced by Allan Pinkerton for the AoP were so off-base that one might wonder whether McClellan wanted them wrong. This nugget actually became the crux of Stephen Sears' interpretation of the McClellan/Pinkerton partnership. Sears weirdly has commited to a "conspiracy/manipulation" view, even though he has admitted repeatedly that McClellan believed in the truth of his own estimates. Now Fishel comes to Sears' rescue with the "smoking gun" of a discovered Pinkerton "conspiracy confession." Here it is, in a letter alluding to an Oct. 4, 1861 report:

[My] estimate was founded upon all information then in my posession, derived from my own operatives, deserters from the Rebel service, "Contrabands," &c, &c; and was made large, as intimated to you at the time, so as to be sure and cover the entire number of the Enemy that our army was to meet.

That was Pinkerton. Here is Fishel:

"Made as to be sure and cover the entire number of the enemy..." In that half sentence we have the explanation of Pinkerton's badly exaggerated estimates that historians have been ridiculing all these years, attributing them to Pinkerton's credulity and general incompetence. Now it develops McClellan was just as aware of the fact as Pinkerton. McClellan's reporting of six figure estimates to his superiors in those early weeks, with little factual basis and before Pinkerton had time to get his feet on the ground, made it necessary for the detective to produce some supporting data. And it was important that the information he provided would ensure against their ever being surprised by enemy numbers they failed to count.

So McClellan and Pinkerton were admitting "making large," and Cox and Sears are spared further embarassment. Or are they? Look at the last sentence in the passage and how it mitigates the sting of the phrase. By Appendix 6, which is devoted to the McClellan/Pinkerton estimates, the idea in the last sentence is developed at length and the phrase "make large" is rendered, in Fishel's judgement, harmless. So the phony confession that renders harm to Pinkerton and McClellan in Chapter 5 is rendered harmless in the back alleys of Appendix 6. A mistake or a connivance?

Note also the conclusions leapt to in Fishel's passage above. Foremost, that McClellan gave out figures "with little factual basis" in those early weeks. Yet we have record of McClellan often bringing sources (spies, deserters, McMundy) around to his chiefs as "factual basis" for his estimates and his letters allude to additional sources (Knoxville agent, etc.). Furthermore, there is an assumption that the detective "had to produce supporting data" for McClellan's pre-existing estimates. Clearly, this is backwards. McClellan had estimates from the War Department, his own spies, governors, runaway slaves, deserters, his own cavalry, telegrams, etc., that formed a picture which Pinkerton's estimates would have to round out. The result would then be McClellan's estimate. In other words, Pinkerton's estimates were not the larger picture, but rather they had to fit into the larger picture. The synthesis of Pinkerton and all other sources of intelligence was McClellan's product, the overall estimate.

The mystery of the Civil War in the East is how these disparate sources could all, consistently produce similar high numbers, the illusion of which had to be squared against Pinkerton's more methodical estimating. For instance, after the Richmond campaign and before Antietam, most or all of McClellan's chief subordinates told Gen. Halleck that they had faced 200,000 rebels on the Peninsula.

At the outset of the Antietam campaign, Stanton's office warned McClellan that there were 100,000 Rebels at Frederick. The New York Herald said there were 150,000. Ft. Monroe's commander passed on two estimates for Frederick: 75,000 and 30,000-50,000. McClellan's cavalry chief told him the figure was 100,000. Penna. Gov. Curtin, whose intelligence organization Fishel praises, warned McClellan that there were 200,000 men with Lee in Maryland and another 250,000 facing Washington. Pinkerton himself estimated 100,000. And Gen. Nathaniel Banks' spies, again praised by Fishel, put Lee's strength at 150,000 in less than a month after Antietam.

From this (which excludes all the high numbers produced by interrogations and spy reports), we get an inkling of where Pinkerton's numbers fit into the pattern of general "noise" surrounding estimates and we can understand Pinkerton's concern about covering "the entire number" of the enemy. But Fishel keeps the noise out of his book. The McClellan/Pinkerton estimates are isolated from context so as to appear freakish figments of two overworked imaginations. The larger mystery of sky-high estimates remains.

Because McClellan and Pinkerton carried off their intelligence files in 1862, and because much intelligence was exchanged conversationally, we cannot know how McClellan figured. Because this particular streetlight is out, Fishel assumes there is nothing to find nearby. This is another mistake. There is much more to say about Fishel's handling of McClellan's estimates, but this can be saved for part of a separate document about the estimates generally.


Perhaps it is not too late to abstract the scheme of the book. It begins with a cursory description of the comparative intelligence assets and products of the two sides at the start of the war. With McClellan's arrival, it reviews his performance in strict accordance with the prevailing Williams/Sears interpretations. McClellan's stewardship is a "travesty" that sets up needed machinery but gets everything wrong. Pope, meanwhile, is a brilliant proponent and user of intelligence who just happens to blind himself by ruining his cavalry in the middle of a campaign and then ignores or misinterprets everything he learns. McClellan's return is followed by an enigmatic Burnside interlude of which we know little because there are so few records. This is followed by a golden awakening under Joe Hooker, whose records we happen to have and who perfects McClellan's systems and machinery and gets everything right. Things work so well that even after Hooker is relieved, Meade reaps the crowning achievment of Hooker's intelligence system, victory at Gettysburg. The book then skips Meade's adventures until Grant arrives, at which point structural changes and dissension cleaves Hooker's legacy and the organization is distributed in bits.

This is coarse shorthand, but takes us to the next point: analyzing the scheme. McClellan and Pope are the only commanders whose relationships with their estimates are thoroughly examined and explained. Burnside represents a mystery to Fishel, at least in respect to what he knew and believed. Meade and Grant are passed over quickly. And Hooker, the centerpiece on Fishel's table, is entirely problematic. First, Fishel fails to connect him convincingly with the estimates and breakthroughs his staff produces: Hooker's use of these is inferred or guessed at. Second, Hooker gave no credit during or after the war to his intelligence workers or their product for decisions he made. Third, he actually denounces his intelligence staff at one point, which Fishel glosses over as inconsequential. Thus, we have a big basket of apples, oranges, peaches, kumquats, all of which the author pretends are comparable, though only McClellan and Pope are really comparable under the "full treatment" offered. And as indicated, the McClellan material is largely missing, making nonsense out of Fishel's guesswork and his associated judgements.

I mentioned Fishel's easy passage over Hooker's potentially book busting problems. This is a trick Fishel performs a second time, with Meade. It is a short bit worth noting.

Meade: The intelligence bureau is "good for nothing." "It furnishes no information not already received through the cavalry."
Fishel: "Since [cavalry commander] Pleasanton's record as an intelligence source made this comparison ludicrous, it appears Meade was quite severely agitated."

Meade didn't mean it? He was having a bad day? Again, we are guessing our way into a mind and substituting the hard fact of a stated judgement with the author's contrary foregone conclusion. This is an amazing way to treat evidence and it happens in a book about the treatment of evidence.


Early in this review it was noted that to mark an estimate as wrong we need data, starting with the "true" numbers. This is not an easy thing for a Civil War author, because numbers and losses are the subject of extensive controversy. An author is compelled, by decency and professionalism, to say, "I take my estimates from ___ because ___." Fishel does not do this. At the end of 700 pages, I cannot tell where his numbers come from except, by implication, from the OR. He does not cite the later authorities Fox or Livermore and two explicit citations he gives come from a secondary sources -- a Sears book -- which itself does not cite a source. We are thus travelling through 24 chapters of "his numbers are wrong, or his numbers are right" without ever learning why Fishel thinks right is right.


The Secret War fulfills a need that makes it a default standard. Hooker specialists may find it interesting. However, this book is a tragic misuse of luck, opportunity, and years of research.


Amazon link

(c) 1998 The McClellan Society