by Bill James
In the years other than the years that The Man from the Train was active, unremarkable family murders will account for almost all of the total. There is the other serial murderer in 1911, and there were a few Italian immigrant families murdered by The Black Hand in the very early part of the century. The 1914 murders at the Wisconsin home of Frank Lloyd Wright, although they cannot be labeled unremarkable, actually meet all of the “normal” conditions that I outlined above. But here’s my point: we are not talking about a large number of crimes that could be sifted and sorted to “find” the murders that fit the pattern; it’s just not that way, at all. We’re actually drawing from a very small pool of family murders. If you remove the unremarkable events, then the cluster of characteristics identifying The Man from the Train dominates the remaining events.
* * *
(3) Let us approach this as a mathematical query, understanding that some of you are not comfortable with story problems and will want to skip ahead a couple of paragraphs. We can begin by distinguishing between the fundamental conditions of the crime, the idiosyncratic behavior of the criminal, and the fetish behavior. The fundamental conditions of the crime are:
• A family is murdered in their beds.
• The family lives within walking distance of the railroad track.
• The crime occurs near the hour of midnight.
• The crime is committed with the blunt side of an axe.
• The crime occurs in or near a small town, usually unincorporated, with little or no regular police presence.
• The crime occurs without any warning, without a robbery, and without any apparent motive.
• It is in no way clear or obvious who has committed the crime.
The second-level stuff . . . breaking into the house through a back window, removing a screen and leaving it propped up against the house, removing the chimney from a lamp and carrying the lamp around the house, picking up the axe from the family’s yard or a neighbor’s yard, leaving the axe at the scene of the crime, washing the axe handle in a bucket of water, committing the crimes on weekends . . . these are idiosyncratic behaviors, as opposed to fundamental conditions of the crime. And the third-level stuff, such as covering mirrors with cloth and paying special attention to the prepubescent female, that’s fetish behavior.
We’re just dealing now with the fundamental conditions of the crime, ignoring the idiosyncratic behavior and the fetishes.
Mathematical query: Given that there are about ten family murders per year in the United States in this era, given that most family murders will occur in daytime, given that most family murders will obviously have been committed by family members or persons close to the family, given that most of those will be committed with a gun or a sharp instrument other than an axe, how many murders would you expect there to be in the United States, per year, in which:
• A family is murdered,
• Without any warning,
• With the blunt side of an axe,
• With no robbery,
• At or near midnight,
• In or near a small town with no police force or very little police protection,
• Within a few hundred feet of a railroad track,
• Without any real evidence as to who has committed the crime?
The mathematical answer is: zero. As a random set of facts, you wouldn’t expect any murders to meet all of those conditions, in a typical year or in a typical five-year period.
You wouldn’t expect to find any murders that meet all of these conditions, at random, if there were a hundred families murdered a year. If there were a hundred families murdered a year in this period, it might happen (at random) that thirty of those would be murdered with an axe, and (let us say, very generously) twenty of them with the blunt side of an axe. Ten of those twenty crimes might occur in small towns with no police presence; three or four of those ten might occur within an hour or two of midnight. One or two of those might occur within a half-mile of a railroad; one might occur near the intersection of multiple railroad lines. That one crime might occur without any warning and without any real indication as to who was responsible, but most likely not. Given that one crime, there might or might not be valuables left in plain sight.
Realistically, there would probably not be a “random match” for this basic set of facts, even if there were a hundred family murders a year in the United States. And, in fact, if you choose a five-year period outside this period we are discussing, let’s say 1922 to 1926, or 1931 to 1935, whatever . . . in that five-year period there probably will not be a single family murder in the United States that matches these fundamental conditions. There might be one or two. The odds against there being a random cluster of such events, with no one criminal causing the cluster, would be astronomical.
But that’s just based on the fundamental facts of the case—not the idiosyncratic or fetishistic behavior. If you factor in the long list of idiosyncratic behaviors, it’s not thousands to one against there being a random cluster of such events; it’s billions to one. That is my argument, in a nutshell: it is simply not reasonable or logical to suppose that these could be random, unconnected events.
In February 1902, six members of the Earl family were murdered with an axe near Welsh, Louisiana; a man named Albert Batson was convicted of the crime and executed for it. The house was right next to the railroad track. But in that case it is obvious that the crime was not committed by The Man from the Train:
1. The crime was committed in daylight.
2. The crime was committed with the sharp side of the axe.
3. A note was left on the door to divert attention from the crime.
4. A team of mules was stolen from the property.
5. An effort was made to sell the mules.
Very obviously not The Man from the Train, and there are other crimes like that. My point is that when it is not him, it tends (most often) to be obvious that it is not him, because a set of facts at random will not match up with the set of facts that identifies our particular nutcase. It might match up at random on a couple of points, such as the railroad track and the axe, but it won’t match up on the longer list of identifying points.
* * *
(4) The crimes are also united by obvious geographic and chronological patterns that could not reasonably be explained as random outcomes.
June 5, 1910; Marshalltown, Iowa
November 20, 1910; Guilford, Missouri
December 7, 1910; Johnson County, Kansas (near Martin City, Missouri)
March 21, 1911; San Antonio, Texas
Take a map, put pushpins in the map 1-2-3-4, and you will see that these four crimes form an obvious geographic unit, with a clear and consistent pattern of movement. The geographic sequence lines up perfectly with the chronological sequence; not only that, but the time gaps are almost perfectly proportional to the spatial gaps.
Almost all of The Man from the Train’s career is geographically and chronologically coherent in this way. Occasionally, such as after the murders in San Antonio, he will take a longer than normal train ride and surface in an entirely different part of the country two or three months later. But then there will be a cluster in the next place, and then a little gap, and then a cluster in the next place. He kills a family in Georgia, then one in Florida, then another one in Florida, then one in South Carolina. The only time in the entire series when the geography doesn’t make sense is the crime in Nova Scotia.
If the crimes were not connected, why would that be true?
* * *
(5) The crimes are united by many different things, other than the broad outlines of the facts and the idiosyncratic/fetishistic behavior of the criminal. To name one of those things: the lumber business.
The majority of The Man from the Train’s crimes were committed in places where the primary industry or one of the two primary industries was logging. Sometimes it’s not logging; sometimes it’s mining. Think about it. What do you use in the logging business? An axe.
What do you use in the mining business? A pickaxe.
Hurley, Virginia, where we started the book. Logging town.
Ardenwald, Oregon. Logging town.
Rainier, Washington. Logging town.
Colorado Springs, Colorado. Logging town.
Ellsworth, Kansas. Mining town.
Occasionally it isn’t a mining or logging town; Villisca wasn’t, and San Antonio wasn’t. Sometimes he is just passing through town.
Anyway, let’s take these three facts:
1. That most of the murders occurred in logging towns,
2. That most of the crimes occurred on the weekend, and
3. That the murderer left money and valuables in plain sight in almost every case, or else the crime was committed against persons who had nothing worth stealing.
When you put facts together and they make a sensible, coherent package, that is an indication that those facts may belong together. If you put those three above facts together, they make a sensible, coherent pattern: The Man from the Train worked for a living. He went to logging and mining towns, because logging and mining camps were always hiring workers. He committed his crimes on the weekend, because he worked during the week. And he didn’t steal things from the murder scenes, because (1) he knew very well that stealing things from the scene of a murder would get you hanged, and (2) he wasn’t desperate for money. Because he worked.
* * *
(6) Another thing that ties these cases together is the sheer, astonishing competence of the murderer. He is so good at what he does that you almost don’t see it; you almost take for granted that this is the way it goes.
In many of the newspaper reports on these crimes, reporters are baffled by how the murderer is able to overcome the entire family so rapidly. In Villisca, for example, tests showed that the staircase creaked loudly, and that it was entirely impossible to climb the stairs to where the first murders were committed without making a loud noise. For a hundred years, people have been puzzled by how the murderer in Villisca was able to get up those stairs without waking the victims—and in San Antonio there is the exact same scenario. The stairs creak, and the police cannot understand how the criminal was able to get up the stairs without waking anybody up.
Well, I understand how he was able to do this. One minute after his first foot hit the staircase, everybody upstairs was dead. Two minutes, tops. They simply did not have time to react.
What The Man from the Train did was not easy. It was enormously difficult. If you believe that the murders were committed in one case by a jealous ex-boyfriend, in another case by a business rival, in another case by a deranged racist, then you will be very puzzled by how exactly he was able to do this without the family waking up. But if you realize that it is one person committing many of these crimes, then it makes perfect sense: he is able to do this because he knows what he is doing. He has done it dozens of times. He’s not tiptoeing timidly up that staircase like a normal person would; he’s charging up the staircase like the breath of Hell.
The astonishing competence of the murderer goes beyond that. He is able to slink into town, commit his murders, spend an hour or two in the house after the crime without being discovered, and get out of the house and out of town before anybody knows he is there—again, and again, and again.
One of the ways that we know that these crimes are a package is that it is not reasonable to believe that a large number of different people possessed the skill and the self-possession necessary to have committed these crimes without getting caught and without letting potential victims get away from them.
* * *
But the main reason that The Man from the Train’s murderous skein went undetected for more than a decade—and a primary reason why the series of crimes went unsolved—is the irrational resistance to believing that the crimes were related, even when there were sufficient facts to allow a clear-thinking person to see that the crimes had to be related. I have to make three arguments at once here, but they tie together. The Man from the Train was shielded from detection by three factors:
1. The fact that the public had no way to connect the dots until the newspaper wire services reached a certain point of maturity.
2. Very unusual things flabbergast us; that is, they stun us into disbelief. This is different from rational skepticism. Certain experiences and certain ideas seem unreal, and we resist believing them because they are so far outside our experience that we cannot process them as being a part of what seems like real life.
3. In many ways, experience is the worst teacher. In many ways experience is the only real teacher, yes, but in certain situations experience will teach us to believe something that is not true. Experience teaches people to believe that the earth is flat. Experience teaches people to believe that their house will not flood because the house has been there for 120 years and has never flooded before.
Experience teaches people to believe, sometimes, that what is usually true must always be true. Experience teaches detectives to believe that people are usually murdered by someone they know, someone in their life.
Unfortunately, the police for a hundred years “knew” this to be true even when it wasn’t true. When police encountered a serial murderer, prior to 1975, they almost always insisted that they were dealing with a series of unrelated crimes. When the Manson family committed one gruesome set of murders one night and another the next, police insisted that the two crime scenes were not related, assigned different investigators to the two crimes, and refused to “join” the two investigations—even though it is difficult to see how the connections between the two crime scenes could have been any more obvious. According to Jeff Guinn in Manson (p. 273):
From the outset, the Tate and LaBianca cases were hampered by the unwillingness of the investigative teams to share information. . . . They often operated out of the same long squad room, but never effectively cooperated. LAPD administration was unconcerned—after all, the cases really had nothing to do with each other.
When the murderer known as BTK committed obviously similar crimes in Wichita, Kansas, in the 1970s, junior police officers were specifically forbidden from speculating that the crimes might be linked—even after BTK wrote to the newspapers to claim credit for the crimes.
That was irrational skepticism, driven by experience. Irrational skepticism of this nature survives and pollutes this discussion to the present day. JD Chandler, in the book Murder & Mayhem in Portland, Oregon (page 87), makes the following quite incredible statement about the links between the murders of the Hill family in Ardenwald and the murders 365 days later of the Moore family in Villisca:
The shocking similarities between the two crimes were probably in the reporter’s imagination because no charges were ever brought against anyone except Nathan Harvey.
That’s it? That’s all it takes to convince you that two crimes are unrelated—no charges are filed?
By that standard, all unsolved crimes are by definition unrelated to one another. The crimes of Jack the Ripper . . . all unrelated. No charges were ever filed, so obviously, there was nothing linking those crimes except the imagination of the reporters. That’s irrational skepticism.
Raymond Hardy was suspected of murdering his family on a farm in Iowa on June 5, 1910. The county sheriff thought Hardy had committed the crime and blackened his name in the newspapers, but he was never prosecuted because there wasn’t a scintilla of evidence to support the sheriff’s slanders. Hardy wasn’t a bad person or a troubled person, and he had no motive; he was just a nice farm boy, supposed to be married a few days later. After he was exonerated he lived a long, normal life, living almost sixty years after the murders. To my eyes, it is as clear as could be that Raymond Hardy was entirely innocent.
But modern articles about that case—and there are several of them on the Web—almost all insinuate that Hardy was guilty, because, well, you know, otherwise we would have to believe the family was a victim of a roving monster, and we can’t believe that.
A
fter Hardy was cleared, a man stepped forward to claim that he had seen a suspicious person, covered in blood, on a train leaving town on the morning after the murders. All or almost all of the modern accounts of the crime will say:
• Hardy was set free in part because of the story about the stranger on the train.
• The local people were anxious to have some alternative explanation for the crime.
Both parts of that are bass-ackwards. The story about the stranger on the train surfaced only after Hardy had been legally exonerated and released. And the entire history of crime screams that people prefer to believe that they know who committed a crime, rather than accepting that the crime may have been committed by some unknown person for some unknown reason. Skepticism in favor of a presumption of guilt when the facts do not support a finding of guilt is irrational skepticism. And irrational skepticism is the smoke screen behind which these murders have been hidden for a hundred years.
* * *
I promised you earlier my list of thirty-three elements that we used to identify crimes that may have been committed by The Man from the Train; they are not thirty-three unique elements, but thirty-three elements with some overlap and redundancy. Here is the list:
1. Extreme proximity to the railroad tracks (usually within a quarter of a mile).
2. Crimes usually occurring near the intersection of two railroad lines.
3. The use of an axe.
4. The use of the blunt side of the axe, using the axe as a bludgeon rather than as a cutting instrument.
5. The taking of the axe from the family’s woodpile or from a neighbor’s woodpile.
6. Leaving the axe at or near the scene of the last murder.
