Speculation in Historical Writing
Bachman, and to a lesser extent, Anderson and Faulring, accuse me of being so speculative in my writing that my book has speculation and little else. Bachman repeats a sarcastic comment by a BYU friend that if the "probability and possibility" words probably, perhaps, may have, might have, must have, undoubtedly, apparent and apparently, seems likely, were taken out of my book, it would be a pamphlet. (p. 108.)
However, he does assert that some speculation language (using the kinds of words his BYU colleague has criticized) can be a valid part of historical writing. So the argument is not simply that I am speculative, but that I am speculative too much. After reading Bachman's treatment here, a friend who had read my book told me that Bachman left the impression that there was no evidence or authentic historical research reflected in my book, but that it was entirely wild speculation and invention.
But if Bachman asserts that some speculation is acceptable, even needed, and that too much speculation is the sign of shoddy history, he does not give us any guidance on exactly how much speculation is allowable, or desirable. How do you define how much "probability language" is acceptable? Bachman gives us no help here. Having labeled my book as based on speculation rather than evidence, he has accomplished his purpose, and moves on.
To consider the question of how much "speculative" language is acceptable, I turned to a biography I happened to be reading recently after a visit to Mount Vernon, John E. Ferling, The First of Men: A Life of George Washington, regarded by many as the leading biography of Washington, and I made a cursory search for "speculative" language. I bold these examples of "speculative" or obviously "interpretive" language for convenience:
"Washington seemed even less eager to spend time with his mother than he was to attend church. A domineering woman, she must have nagged at her son when they were together." [Later in the same paragraph:] "He apparently did not invite her to stay at Mount Vernon." (p. 76)
So we might ask whether this is a bad paragraph because it has an abundance of speculative words in it. Or was the clustering of "probability" language appropriate for the particular subject addressed?
In another sequence of paragraphs, Ferling writes: "Washington had never met the venerated Franklin. They apparently got on well enough, but the old sage, not unlike Washington, was surprisingly reserved, and the two men must have found their attempts at light conversation to be awkward. The commander had only a passing acquaintance with Lynch, having met him when they both served in Congress. But he must have been delighted at the sight of Harrison, not only a friend, but a Virginian who had just returned from a vacation in that province." (pp. 133-34)
In the next paragraph: "General Washington must have blanched at the first directive. [Pay of officers must be reduced, according to a congressional committee.] Whatever his feelings about the wisdom of an attack, the notion of civilians four hundred miles removed from Boston dictating strategy to the army had to be galling." Then there is a very colored, interpretive word at the end of this same paragraph: "Their bluff called, the congressmen backed off, and during the final five days of the conference they obsequiously accepted almost every urgent request made by General Washington." Compare the word "domineering," a judgment word, used previously. Again we must ask, are these passages overrun with speculation evidence that Ferling is a bad scholar?
Bachman himself frequently uses "probability language." On p. 133 (again, I bold for convenience): "He [Compton] too may have been persuaded by doctrines such as those of kindred spirits, free love, and spiritual wifery. In 1845 William was probably theologically closer to John C. Bennett than Joseph Smith in his thinking and acting in relationship to plural marriage." And again, on p. 127, Bachman writes, "Apparently he believes that the four sources cited establish the relationship." Again on p. 131, "It appears that Compton has taken the bait." On p. 129, "It appears that the doctrine of 'kindred spirits' is Hall's." We can ask again whether these are bad paragraphs due to the clustering of probability language.
My answer to the above questions, of course, is that Bachman and Ferling are not bad scholars because they used probability language, and those are not bad paragraphs, merely because they contain probability language.
Here is how I view this issue. Many non-historians have the point of view that a historian merely "reproduces the facts." This is a basic misapprehension about history. Even a historian's reproduction of facts reveals his or her point of view, as he or she selects which facts, out of an infinite number of facts, he or she will refer to or quote. What every historian does is select and present evidence, much of which is fairly undisputed (though, as I mentioned above, no evidence is perfect), then he or she draws interpretive conclusions from the evidence. All historians do this. It is the essence of their job. They are not merely reprinting phone books or statistical facts. All historians have a point of view, opinions, predisposition (though, as in my case, their exposure to evidence will cause them to develop and revise opinions). So Bachman, Anderson and Faulring's branding me as "speculative" is odd; they are in essence saying, if you do not have exactly their biases, you are "biased." Bachman critiques me for having "opinion" (p. 136) -- as if he does not have strong opinions himself. Good historians' opinions are certainly guided by factual evidence, and they make an honest effort to not let their biases shape their view of evidence, but opinion and individual perspective are always there. Probability language is closely aligned with interpretive history.
However, I certainly agree that the speculative, interpretive aspect of history can fall short of absolute truth. (No historian, of course, achieves absolute truth.) Interpretation in history can go astray in a number of ways.
First, an author may make an argument, write an article or book without sufficient research. If there are six relevant pieces of evidence, and the author uses only two of them, his or her argument may well be flawed. I don't think that Anderson and Faulring and Bachman believe this of me -- in fact they almost criticize me for using too many sources.
Second, an author may adduce the facts adequately, then make an incorrect construction from them. This can be done innocently. An author collects three facts, then makes a judgment on what probably happened. However, three more facts may arise that put a different light on the circumstances, and the historian may either change his mind completely, or at least continue his previous reconstruction of the past with modifications. Sometimes a historian's extreme, irrational bias can produce consistent illogical jumps that do violence to the evidence. Arguments of white supremacists, arguing that Jesus was not a Jew for instance, present good examples of this.
My main point here is that all history is interpretive and speculative, though the best history works from evidence collected and adduced as skillfully and honestly as possible. Any attempt to portray my book as speculation only, not based on authentic evidence (Bachman refers to my writing as "pure fantasy" at one point, p. 133, as if I were not using evidence at all), is obviously ludicrous. I constantly refer to primary evidence, diaries, autobiographies, letters, early church periodicals, newspaper accounts, genealogical information. These are quoted virtually everywhere in my book, but the concentrated, technical references to sources are in 623-771, about a hundred and fifty pages of bibliographical references and citations. I have always done everything possible to find the most primary evidence on a subject, then to weigh pieces of evidence judiciously, trying to work in the highly charged emotional battleground of religious history. As early as my Fawn Brodie article, I was concerned with the challenge of interpreting evidence in Mormon history, and critiqued Brodie for relying too much on secondhand, printed, anti-Mormon sources.
So any responsible, real history will have a great deal of interpretation, which by definition deals with speculation and probability, though it will be based on evidence.
Is there such a thing as too much speculation? For instance, do the paragraphs by Ferling above have too much speculation packed into a small space? I don't know. In my view, you can have too much speculation based on too few facts (or on no facts). But note that Ferling is not simply spinning a tale; he is arguing for a certain historical reconstruction from facts, which admittedly he has marshaled in a certain pattern. He is not arguing from a vacuum. He allows the reader to see the facts he uses, then he clearly separates those facts from his conclusions (using "probability language"), but shows that the conclusions come from the facts.
This is an exciting part of history. You look at the facts, and they lead you to reconstruct a picture beyond what is strictly in the lone pieces of evidence taken one by one. This is not irresponsible history. This is what history does. If a historian becomes immersed in the diaries and autobiographies of the battle of Gettysburg survivors, supported by a careful mapping of the Gettysburg site, along with archaeological findings, he can come to see a vision of the battle. And the overall vision goes far beyond what is found in any one journal or any one bullet found in a hill.
Note also that, for readers or historians who have an extremely rosy view of Washington, they might not be comfortable with Ferling portraying him treating his mother distantly. So they might accuse Ferling of biased, negative, speculation. Thus we may ask if Ferling was attacking Washington vindictively, or whether he felt that the evidence authentically pointed to a problematic relationship between Washington and his mother, and Ferling felt he should discuss it, because it was there. (And because mother-son relationships are significant.) Was he intent on debunking, or was he merely giving a balanced portrayal of Washington as an individual who at times had problematic relationships? (And most or all humans have some problematic relationships.) It is possible that we have a case where Ferling felt a personal dislike for Washington, so overemphasized innocent facts to create an unfairly negative picture. But it is also possible that Washington has been intensely idealized as the first American president, and when actual, normal human problems appear in his life, in the historical record, the historian may responsibly reflect them without vindictiveness.
Another distinguished book of history, Laurel Thatcher Ulrich's Pulitzer-Prize-winning A Midwife's Tale, is worth considering as we discuss "speculation" in history. In this book, the main document is a diary of midwife Martha Ballard. Ulrich quotes it day by day, then stops and makes long, imaginative reconstructions of what happened on those days. Certainly, Ulrich makes educated guesses, based on voluminous research. But everything in this wonderful book, outside of the diaries themselves, is reconstruction, speculation -- possible or probable reconstructions, in fact extremely sophisticated, brilliant reconstructions, but going well beyond the flat record found in the diary. So should responsible critics point out that there too much "speculation" in Ulrich's book? Is it fatally flawed because it is "filled with speculation"? Or should she have merely printed the diary itself and avoided injecting her personal reconstruction and judgment? Here, certainly, you could say that without speculative reconstruction, the book would be a pamphlet.
My view is that what we should be concerned with is not whether historians use interpretive "probability language," but whether they are consistently illogical in their conclusions and are making wild, irrational fantastic jumps in their possible and probable reconstructions.
Let's look at Ferling's "probability language" again: "Washington seemed even less eager to spend time with his mother than he was to attend church. A domineering woman, she must have nagged at her son when they were together." Later in the same paragraph: "He apparently did not invite her to stay at Mount Vernon."
Now I will take that "probability language" out: "Washington was even less eager to spend time with his mother than he was to attend church. A domineering woman, she  nagged at her son when they were together . . . He  did not invite her to stay at Mount Vernon."
How do we feel about the second paragraph as opposed to the first paragraph? The tone of the second paragraph is dogmatic. It seems to assert that the author is using certain evidence. It does not complicate the picture with "probability language." In contrast, the tone of the first paragraph is cautious. It asserts that the author is working from evidence, then making an educated reconstruction of what the evidence points to. Using the "probability language" actually shows the author's honesty -- here is my evidence, he or she says, and now here is my argument from the evidence. Probability language is a safeguard for intelligent readers. It allows them to quickly assess the evidence and make a judgment. (If they want to pursue the matter further, they can go to the original documents and check context for themselves.)
So, at the risk of seeing myself too sympathetically, I propose that my "probability language" actually reflects my caution and conservatism. I believe this is where Brodie sometimes got into trouble, when she expressed her interpretations dogmatically, as if they were absolute truth, instead of cautiously. (Her friend Dale Morgan criticized her for this.) Certainly, some of my reconstructions may be modified by future research, or even rejected, eventually. But to portray me as a bad historian because I use interpretive language "too much" I think shows a basic misunderstanding of what a historian's task is.
At the risk of beating this issue into the ground, I will briefly look at two of my uses of "probability language." When Fanny Young Murray died, I write, "No doubt her passing was mourned by Brigham . . ." I also argue that Fanny's sister wives and close friends attended her funeral, and were present when she was buried. I had no direct evidence, but it seemed reasonable and probable that such was the case. Brigham Young was her brother. Fanny had lived in the Lion House with these women. This is not bizarre speculation; it is common sense.
When Vilate Kimball died, I write that Helen Mar "undoubtedly tended her as she gradually lost her faculties." (In Sacred Loneliness, 514.) Now, I did not have a shred of evidence documenting Helen at her mother's bedside. Yet I felt that it was extremely probable that an only daughter who dearly loved her mother, and who lived a half a block away from her, would take time tending her mother as she approached death. I enjoyed visualizing that very probable scene.
Is there really anything sinister or wildly speculative about such an interpretation as this?
Incidentally, a syntactical technique that Bachman uses repeatedly that is an extreme form of probability language is the rhetorical question -- for instance, "Did his presuppositions shape the study as he sought evidence to validate them?" (p. 118.) Such a rhetorical question shouts "Yes" as an answer with every word. It is ironic that Bachman would criticize my cautious use of probability language, then repeatedly use the rhetorical question as a form of speculative innuendo -- the statement just quoted is based on no facts whatsoever. Bachman did not know me before I began researching my book; he knew me at no time during the writing of the book; he met me briefly after the book was published. Furthermore, as I have mentioned, I had no strong preconceptions about polygamy before I began researching my book.
Another critique of Bachman, and Anderson and Faulring, that is odd is their charge that I take evidence from "disparate" sources. (See Bachman at 121, "patchwork," 135, "disparate authorities scattered over decades"; Anderson and Faulring at 69). It is a standard in the practice of history that any kind of evidence relating to a phenomenon can be considered, and in fact, should be considered. For instance, if I have an autobiography of a woman, and she gives her birth date, but tells us little about her family, it is extremely useful to find primary genealogical records for her. First, it can confirm her memory of her birth date (sometimes, surprisingly, people remember their birth date incorrectly); second, it can give added information on her family situation when growing up -- whether her parents were middle-aged or young when she was born, whether she was the oldest child in a large family, the baby in a large family, or whether she was an only child. The fact that I have combined two disparate sources strengthens my portrayal of this woman. What if I refer to a newspaper story that mentions my subject, and combine that with what she wrote in a diary on the same day? This will again strengthen my case, give more depth to the story. The more pieces of evidence that deal with an event, the better off the historian is. It does not matter in the least if they come from disparate sources. In fact, disparate sources increase the validity of the case.
Anderson and Faulring criticize me for using sources that are contradictory (p. 69, "gross contradictions"). However, once again, I find this criticism hard to understand. This is the common lot of all history. Find two sources and chances are they will disagree on something. If disparate sources contradict each other, this is still extremely valuable, because it leads us to try to examine and resolve the contradiction. (Which can often be done.) The idea that one should simplify by using only one of those sources is not tenable in any way. (Unless some other factor leads us to believe that one of the sources is completely wrong. Even then it is valuable to try to understand how this warping came about, what forces created it.)