Probability that Mary was a murderess

[mnyquist]

In trying to convince the readers of my book that the Mary therein was truly a murderess, I had wanted to additionally employ a probability analysis, but failed in this endeavor. Mary used several techniques, it appears, and had at least one accomplice. Of a very successful mining family, she had unlimited access to money and manpower. Cyanide was also extensively used in mining operations at the time. Poisoning was often misdiagnosed as pneumonia, tuberculosis, jaundice, liver failure, chest pains, apoplexy, paralysis, and gastro problems.

Here are the factors:

1. A total of 15 men earned the wrath of Mary. A result was that 13 of them died under questionable circumstances. Some were sick several times before they succumbed.
2. One was hit, on a bridge, by a train, while working on switch maintenance (no switches on bridges!) on a one-person velocipede. At the time, the fatality rate for railroad workers was 2.67/1000. I suspect he was incapacitated and merely placed on the bridge, close (within 1 mile) to where Mary was living at the time.
3. Two died in mine "accidents," at a time when the fatality rate was 3.13/1000.
4. Eight died of "gastric" or other complaints which could be related to poisoning, at a time when the fatality rate was 1.4/1000 gastro; 2.0/1000 pneumonia; 1.9/1000 tuberculosis; 1.37/1000 heart (about 1.7/1000 average).
5. Two died of unknown causes. Of these, one was a rival banker just in the process of having a new bank built across the street, and the other was an errant husband who was touring the West Coast by car with friends.
6. Two remained extremely diligent, and survived. But, for one of these two, who was living alone was his widowed mother, the mother became paralyzed, dying seven years later. I can find no other instance of Mary targeting a woman, and surmise the real target was the son. For the other survivor, he lived with his brother, who had one of the above mine accidents.

So, how to frame this? In the simplest terms, 13 out of 15 "enemies" had fatalities.

But I am looking for the proof that these were not mere coincidences. Probably, in STAT terms, the probability that all of this occurred. I think this is something like (2.67/1000) * (3.13/1000) * 3.13/1000) * (1.7/1000)......Correct? But these are annual rates, so 2.67/1000 * (years of average employment?)

I would greatly appreciate any assistance you could give. I'm an ex-engineer, but I'm lost. Thank you and best regards, Mike

 

[JeffM]

I don’t think this is going to work. I will venture a bet that all fifteen of her enemies had fatalities, which means they died. If death of a high percentage of one’s enemies is evidence of murder, we all shall end up incarcerated.

You are correct that annual rates of death are not appropriate. If the aggregate death rate due to mine accidents is 3 per year per 1000 employed per year and someone was employed as a miner for 10 years, the probability of dying by accident is closer to 3% rather than 0.3%. You do not say what the base is of your percentages of death rates by type of disease. If that is the number of deaths from that disease by year per thousand of population, that is an almost meaningless number. The probability of an eighty year old obese man dying of a heart attack has only the most tenuous link to the ratio of deaths by heart attack in one year relative to the overall population.

You place a great deal of weight on misdiagnosis without any quantification of the rate of misdiagnosis. Were 90% of those diagnosed of dying of gastric complaints (which probably meant stomach and bowel cancer) really poisoned? That would mean a very high rate of successful poisonings. Which is more plausible: that most diagnoses of cause of death were correct or that most were misdiagnoses of cyanide poisoning.

Did the errant husband die in unusual circumstances? If the errant husband was not Mary’s husband, there is a more probable suspect, possibly several more probable suspects.

You are going to need to show that Mary was a murderer through the standard means: motive, means, and opportunity. And you seem to be making the common mistake of assuming what is to be proved. X was not poisoned, but X’s mother died. Therefore Mary must have poisoned the mother by mistake. That is not evidence. Even if you have persuasive evidence that Mary poisoned specific people, you cannot ascribe to her deaths of other people where there is no evidence.

I am particularly fascinated by the “velocipede.” I suspect what was meant were those muscle-powered devices that allowed railway workers to travel along the rails from one work station to another. If so, a wreck between one of those and a locomotive would likely be fatal only on a bridge because otherwise anyone sane and sober would jump off.

 

[mnyquist]

Of course, everyone dies. But we don't know each other, so I'll cut you some slack! And Jeff, please, I really could use some help.

In the simplest of terms, Mary had 15 enemies. Of these, 13 did not die of old age or natural causes, but of some mishap. About 87% of them. OK, but she had her reasons.
I had two main sources for the fatality statistics:
[1] https://eh.net/encyclopedia/history-of-workplace-safety-in-the-united-states-1880-1970-2/. This is per worker per year, not total population.
[2] https://www.statista.com/statistics/235703/major-causes-of-death-in-the-us/
The book does go into detail regarding motive, means, and opportunity.
Let's take a look at the velocipede accident. There is no need for a signal repairman to be on a bridge except for transit, and his dispatcher would be seriously negligent to have him on a bridge when a train was scheduled to pass. The bridge in question was near a station, so the locomotive was likely traveling at less than 15 mph, in those days. He would have heard it far in advance (or the train could have whistled), and he could have traveled off the bridge by velocipede or foot, or jumped off, or jumped onto the cowcatcher, if he were not somehow incapacitated beforehand.
Regarding the husband, he had been unfaithful, resulting in an illegitimate birth. Then he got sick. He tried to start a new life at a new location, and sent for his wife and son to meet him. As soon as they arrived by train, he again got very sick. Feeling better some time later, they traveled. When they arrived at the destination of their new home, he got sick for a third time, dying four weeks later. The wife and son immediately went back to their old house.
So, back to the statistics. I have two brothers, both dying in mine accidents. 3.13/1000 per year. In those days, you could probably mine for an average of 40 years. so, 3.13 x 40/1000, say, at 8% chance (wow!) a single miner would die in an accident the course of his career. The chance that they both die? I believe .08 x .08 = .0064. Less than 1%.
But we are just getting started. Plus the chance of the train accident? 2.67/1000 x 40 = .11. So, for all three accidents? .0064 x .11 = .00068. Less than 0.1%
Now, let's look merely the chance that a person in those days died of a broadly gastric complaint. Maybe 1 out of 2? This happened eight times. (1/2) to 8th power. .0039, or 0.3% chance.

I was just looking to probability and statistics to lend further credence to my conclusions. Hope I can get some help on this. Thanks to all, and best regards.

 

[JeffM]

The problem with the individual statistics is that unlikely events happen all the time. The probability that one person will win the lottery is minuscule, but someone does. You cannot use the fact that person A won it to argue that person A had the fix in.

Your best bet for using probability is much simpler. See if you can find a standard life table for the period in question. (If you cannot, use a current one. It will not be accurate, but it will be biased against your hypothesis.) Calculate the probability that a person would die no later than that age. Then calculate the joint probability that 13 unrelated people would all die prematurely. It will undoubtedly be a small probability. Thus, you can make an inference from Bayes Theorem that they were not unrelated.

In terms of data quality, you also need to look at the specific incidents to see if they should be excluded from the sample. For example, if a mine accident killed 30 people in a mine where 45 were working, it makes no sense to look at a national average. It is possible that someone could sabotage a mine for the purpose of possibly killing one person, but that calls for considerable knowledge and opportunity. Folks like the Coal and Iron Police were on the lookout for sabotage; you did not just wander into a mine and set charges. On the other hand, if the boiler blew on a donkey engine and killed the driver, it is not that hard to tamper with a safety valve on a steam boiler.

 

[mnyquist]

Jeff: thanks for the responses!
Anyone else want to jump in here?
Yup, this is my problem. Just because someone, for instance, died in a car accident and another by lightning doesn't of itself mean anything.
But I see several patterns here and, admittedly, have not indicated herein all that I know.
It would have been nice to see some feed back on the train accident, given the additional info above. I also didn't mention that, at the time of the accident, Mary was living a mile away. And how to account for Mary's husband's 3 sicknesses, all at critical moments of his life?
I don't really believe in coincidence, over and over.
Regarding the two brother miners, they were killed in separate, single person accidents. One was crushed between a car and a cart in a mine, the other fell down an ore chute in a different mine. Mines where Mary had a ownership interest. Clumsy.
A newspaper discussing one of the gastro deaths indicates that a year earlier, the person in question suffered from severe maladies, including poisoning, and with many symptoms correlating with poisoning.
The probability based on age at death will not necessarily help me, as the ages spanned from 20 to 70. This is why I was trying to quantify the causes rather than age of death. A nephew said the Bayes Theorem is the way to go, but I still can't determine how to set this up. I need help!

 

[JeffM]

Jeff: thanks for the responses!
Anyone else want to jump in here?
Yup, this is my problem. Just because someone, for instance, died in a car accident and another by lightning doesn't of itself mean anything.
But I see several patterns here and, admittedly, have not indicated herein all that I know.
It would have been nice to see some feed back on the train accident, given the additional info above. I also didn't mention that, at the time of the accident, Mary was living a mile away. And how to account for Mary's husband's 3 sicknesses, all at critical moments of his life?
I don't really believe in coincidence, over and over.
Regarding the two brother miners, they were killed in separate, single person accidents. One was crushed between a car and a cart in a mine, the other fell down an ore chute in a different mine. Mines where Mary had a ownership interest. Clumsy.
A newspaper discussing one of the gastro deaths indicates that a year earlier, the person in question suffered from severe maladies, including poisoning, and with many symptoms correlating with poisoning.
The probability based on age at death will not necessarily help me, as the ages spanned from 20 to 70. This is why I was trying to quantify the causes rather than age of death. A nephew said the Bayes Theorem is the way to go, but I still can't determine how to set this up. I need help!

I'll think about Bayes Theorem.

 

[Cubist]

I've been watching this thread. I find it interesting. It sounds like it could be a nice book to read. Although I'd recommend that you write it from the angle of an interesting set of potentially related deaths set in recent history. Give detail about the people involved, their lives, and the different living/ working conditions at the time. I wouldn't recommend that you point the finger too strongly, ie. be objective. But, I'm not an author, so please take this advice with a pinch of salt!

It seems very likely, because she had ownership interest in some mines, that Mary would be a very assertive, confident and outspoken woman with some resource? (Good for her!) Therefore...

• Are you sure that she only had 15 enemies? It doesn't sound a lot to me. And, it doesn't surprise me that some of her dead enemies worked in the mines that she had interest in - I'd actually expect that to be the case. Why would a random miner, working elsewhere, be her enemy?
• She might be very capable of applying regular stress to her enemies, with sharp criticism etc. Depending on the recipient's character and sensitivity, this could have a strong "nocebo", or anti-placibo, effect on their health making them more susceptible to disease. This could also cause a lack of concentration while they are at work. Depression and anxiety was not acknowledged/ treated properly in society back then.

It would be very difficult to factor all these things into an equation to come up with a single probability number. Why not let your readers form their own opinions, while you provide the facts and perhaps a moderate amount of conjecture?

 

[mnyquist]

Hello Cubist:

Thanks for joining this thread!

Actually, the book is already published, since June 9th. It is available on Amazon Books. It has a long title: "Bent on Revenge: Murderous Millionaire Mary: A Genealogical Investigation Goes Wrong," by Michael S. Nyquist.

Many of the victims were my relatives. As such, it is often difficult to control my emotions and remain subjective. But I think I've done the best I can, considering.

I was merely trying to solve an age-old mystery as to who was my true great-grandfather, and the story developed from there into something truly horrific. I couldn't wait to publish this, even though I am still learning some new facts. I am in my 70s and wanted to make sure this story got out. The crimes date back to starting 138 years ago, with nothing ever published about them.

Mary's first husband (who, by the way, she killed (sorry: can't help it!)) is in fact my great-grandfather. He is viewed in heroic terms in at least one part of the US (probably more: eg: former US Congressman), but, like all "heroes," had some flaws. Not quite of the same low standard as his wife, though.

Such a horrific story: it is hard to believe it. That is why I wanted to try to employ statistical techniques to add further credence to it. In solving the mysteries, I had to use genealogical tools, DNA analyses, facial recognition software, and perform extensive research.

She may indeed have had more than 15 enemies. These were merely the ones that made sense to me, falling into three categories of antagonists. Too many coincidences.

Many of your suggestions do feature in the book, including the development of the biographies. Regarding her personality, I am not a psychiatrist, but I believe she stumbled, unexpectedly, into a world she into which she was ill-prepared. Given her position of exceptional advantage, she seems to have been incapable of acting in any documented positive and charitable way, with an intent towards self-improvement. Rather, she was always looking over her shoulder, ready to attack anyone who did not give her her "entitled" respect.

Well, joining this forum and creating this thread has not resulted in the probability formula I was seeking, but, nevertheless, it has been interesting. Still wish I could derive some statistical analysis of this, however. Will keep trying. All the best! Mike

 

[Cubist]

Thanks for revealing more about the story! And let me offer my condolences that your family has suffered so many untimely deaths. As you've seen I'm a little reluctant to suggest probabilities for this question, because of the complicated situation. Once again, it can't easily be captured into an equation. A simplified equation wouldn't give you an accurate/ representative answer. BUT, here's a VERY SIMPLE calculation anyway...

If we look at a period of 40 years AND the probability of a death in every one of those years is 3.13/1000, then the probability of death occurring within those 40 years is...

1 - (1 - 3.13/1000)^40 ≈ 0.11785 or 11.785%. Mining was a risky profession!

OK, if there are 15 such miners, then what is probability that 13 or more of them would die using the 11.785% chance figure

[math]\sum_{n=13}^{15}\binom{15}{n} \times 0.11785^{n} \times (1 - 0.11785)^{(15-n)} ≈ 0..00000000007044[/math]
That's a very low probability indeed. And I accept that only two were miners, so the true figure could be a lower.

To put this in perspective, the world population in 1970 was 3,683,000,000. Assuming that everyone in the world knew a group of 15 young people who worked in the mines (silly I know), then how many people would we expect to experience 13 of their group dying during the next 40 year span...

0.00000000007044 * 3683000000 ≈ 0.26, or about a quarter of a person might experience this!

--

But please don't read too much into this. As has already been said, this doesn't capture the full situation. If I wanted to kill someone I couldn't imagine how I'd go about staging it to seem like a railway or mining accident. But perhaps she was clever and evil enough to pull this off. And what are the chances that someone who did this could actually get away with it so many times without being caught or leaving any concrete evidence (which isn't taken into account above). I don't know.

In many ways this reminds me of the Drake equation, which attempts to calculate the number of alien civilizations in our galaxy that we could potentially communicate with. Why? Because the inputs to that equation keep changing as our knowledge increases, and this changes the output. Also people have suggested modifications to the equation to improve its accuracy. But does this equation give us anything near to the truth? And will we ever know the true number of such civilizations .

 

[mnyquist]

I like this equation. Now we're getting somewhere, but with the cautions you mentioned. Also, two dots after the 0...think you meant one.
However, of course:
Only two died in mine accidents. One died in a train accident. The others had gastric "disorders." This greatly complicates the equation, I am sure. Would the equation basically hold if the rate for train fatalities, and for "gastric," were slightly altered (increased) to match the rate for mine accidents? As per earlier, they are all between 1.4 and 2.1/1000/year. Then, could I use your equation? I could say, 13 had various accidents, the rate of which for each was approx. 3.13/1000/year? Your result shows how unlikely these events would have been. Or by converting to various accidents, do I have to increase the rate to the total for all such deaths, ie- (3.13 + 2.67 + ....)/1000 x 40? The result would still be improbable, I believe.
As for how it was accomplished. Mary had 1000s of oft-rough, rugged and poor men working for her, and I don't doubt she could have found someone(s) amongst them who could arrange for those two types of mining accidents for a fee. Ditto for the gastro episodes. One possible poisoner, as per the book, seems to have been rewarded years later with a plum position. I found nothing else anywhere about this man's life.
Mary's empire was far-reaching, and the deaths occurred in many locations. She had mines, hotels, shipping, and other businesses in Michigan, Minnesota, Colorado, Arizona, New Mexico, Mexico, Illinois, Ohio, and New York, at least. No policing agencies in those days (even in these days!) would have been able to coordinate and piece all of the distant events together. For the more local events, don't forget she was the wife of a former US Congressman (who had also been a member of the Mich House & Senate, and former local Mayor), and therefore would have been generally above suspicion. In fact, I believe these are two of the things she was relying on: the lack of coordination; and being above suspicion. I think she came close to being caught at least thrice, when inquests and/or internal enquiries resulted.

 

[Cubist]

Hmmm, how to refine the simple method of post#9?

Obviously, the result obtained is going to depend on the question asked. We need to be careful.

To obtain a fair result, I believe we shouldn't be too specific. I think it might misrepresent things if we calculate the probability of:- 1 railway worker dying at work; 2 mine workers dying at work; 8 people dying from a "gastric" disease and 2 people dying from an unknown cause FROM a group of 15 given that 1 works on the railway and 2 of them work in a mine. (See * below)

I'm not sure what @JeffM thinks, but I'm starting to think that we should be looking at each individual's age at their time of death. What's the probability that, in a group of 15, 13 of them died between the ages {4 to "age of death#1", 4 to "age of death#2", 4 to "age of death#3", ... , 4 to "age of death#13"}. I thought we could start at 4 because there seems to be a slight (kick the ) bucket curve so that there's an increased chance of death for the very young, but if you make it to 4 then there's a reasonable chance that you'll live a long life. See the data in this link https://www.ssa.gov/oact/STATS/table4c6.html .

Or perhaps we could consider the 3 work deaths separately. I'm not sure. I'd be glad to hear from other helpers.


*- if a question is too specific then the result will always be a small probability. For example, consider shuffling a deck of 52 playing cards. Draw ten cards and lay them face up on the table in the order drawn. You think, "can I get that result again?". You start to repeat the process once every 15 seconds. How long would we expect before we get the same same result? I think that the answer is 27 billion years. To put this in perspective the universe is estimated to be about 14 billion years old. Should we feel special that we had that particular result? No. We were always going to obtain some list of 10 cards.

 

[Dr.Peterson]

*- if a question is too specific then the result will always be a small probability. For example, consider shuffling a deck of 52 playing cards. Draw ten cards and lay them face up on the table in the order drawn. You think, "can I get that result again?". You start to repeat the process once every 15 seconds. How long would we expect before we get the same same result? I think that the answer is 27 billion years. To put this in perspective the universe is estimated to be about 14 billion years old. Should we feel special that we had that particular result? No. We were always going to obtain some list of 10 cards.

This is my concern about the whole question. We don't know what denominator to use (what population to consider -- just these 15 known "enemies", or everyone she came into contact with, or something in between), or what numerator to use (only these particular means of death for these individuals, or anything bad happening to the person at all). In general, after-the-fact probabilities are always prone to this error; see Should Rare Events Surprise Us? When something surprising happens, and we want to decide whether it's unusual enough to suspect some crime or trick, we need to consider all the other events that would have been equally surprising, and all the other things that didn't happen, not just the one specific event.

So if you do any calculation at all, in my mind it has to be rather broad, such as having at least 15 out of N people with any negative interaction with her, all dying in some specified time period, from any cause.

It's also important to remember that correlation does not imply causality; maybe someone else in the lives of all these people killed them, perhaps even to frame your suspect. Probability arguments are not admissible in court, I imagine, except in particular cases (like fingerprinting), and when used in law enforcement would only lead to suspects, not proof. But I'm sure we all know that. It's the actual evidence that counts.

 

[mnyquist]

Welcome, Dr. Peterson!
Well, I'm glad that it is such a difficult problem to frame, since it had defeated me.
A couple of things:
I hadn't commented on the "Universe" and life issue. But I don't think I have a continually evolving set of facts such as in that case.
Mary really didn't care how old you were when you earned her wrath. As this wasn't an issue for her, any attempt at analysis in terms of age is bound or at least likely to fail. Just out of interest, I took a look, and the ages went from 20 to 78, with an average of 49.8. This is close to the life expectancy at the time; lower if we take out early childhood deaths, as suggested.
The problem remains with quantifying the three types of "ailments" into some sort of equation.
If I try to frame this in another way, I could do it by type of victim. However, once again, they fall into three categories: members of the "Z" extended family (#7); errant husbands and close family thereof (#3), and rival bankers (#3). With three categories, I believe I have the same problem as with cause of death. The death rate for "Z" brothers and fathers was 3/4; "Z" spouse's brothers 2/3; husbands 2/2; bankers 3/3.

13/15, or 87%. I was aware of the pitfalls with statistical methods. More so nowadays, when statistics are unfortunately used deviously to push a particular agenda. I was looking for a reasonably defendable equation. I was also looking for a result which yielded a 99% probability, or greater, that Mary was involved. To me, that would be close enough to support my hypothesis. But maybe 87% is also close enough, at least for me.
So long ago. I have come close, but agree I can't prove anything. But when I take the book in its entirety, the circumstantial evidence reveals several patterns. And the self-interest at all costs continued for subsequent generations.
No one was looking at Mary. In those days, there had been no such an entity as a female serial killer in US history. Thus, unlike what has been suggested above, no one was killing in an attempt to frame Mary. The first US female serial killer was Lizzy Borden, who was a contemporary of Mary.
As for the time frame, this spanned several years. I think it was like a hobby. There was a period where it appeared her activities went dormant. I think two things happened here: she could not track down who was left of "Z", and things were going very well for her. Then a second stage occurred when certain tings went sour for her again.
Thanks for the posts! I am very grateful. Mike

 

[JeffM]

The relevant time for each person is from the time that THAT person incurred Mary's wrath to the time when THAT person died. The times would be different for each one.

I am still chiefly bothered by two things. One is the presumption that the cause of deaths (other than the apparent industrial accidents) were all misdiagnosed as natural when they were poisonings. It seems to me that the whole argument relies on the assumption that the cause of death was misdiagnosed 12 times. Let's say the probability of misdiagnosis from long-term cyanide poisoning was very high, say 80 to 90%. The probability of 12 being misdiagnosed is then about 7% to 28%. That is not a persuasive statistic.

The other thing that bothers me is that what seems to be hypothesized is long-term poisoning that was not detected and did not affect others in the household. Mike addressed that in a way with the mother who died and was paralyzed, which is indeed a symptom of low-dose cyanide poisoning. That moves us from 12 to 13 misdiagnosed deaths.

But actually the two sets of probabilities need to be combined. Long term poisoning that was not detected and materially affected only one person in the household, Plus death by poisoning that was misdiagnosed. If we take both of those as being quite probable at 90% each, we are at a probability for thirteen cases of just under 7%. Those 90% statistics are of course speculation. They strike me as high. At 80%, we drop to 3%. At 50%, we are effectively at 0. In short, probabilistically, the argument is very sensitive to assumptions about probabilities, not surprising when we are dealing with an exponential function. To get a probability of 50% or more, we need the two probabilities to have a geometric mean of about 97.5%.

 

[mnyquist]

Dear All:
Sorry it has taken me so long, but I thought of another plan of attack. Kindly let me know if this is more amenable to a statistical analysis.
I must have been thinking similar to Jim, for I had already embarked on this!

I have taken a look at the timeliness of the deaths and illnesses. Note also that there was a long lull in the action, which accounts for the late actions. Many of the "infractions" resulted in death or injury shortly thereafter. Here is what I have:

Husband 1. Illegitimate child with Z in 1884. Moved Z, found a spouse. Became sick 1886.
Husband 1. Tried to moved family out-of-state 1887. Sick immediately after 1st portion of move. Sick immediately after 2nd portion of move, dead after 4 weeks. All three times gastro.
Brother-in-law of Z. Died age 23, 3 years after illegitimate child. Consumption. [Note: Z had been forced into a sham marriage and relocation]
Rival Banker 1. Started building new bank across from Mary's bank in 1889 with banker 2, died same year. Cause unknown.
Brother of Z. Died in train accident 7 years after birth of child. Z had been forced to move in 1883. It is assumed that the brother was starting to learn, years later, of the true story, and was starting to repeat it. He was 27 years old.
2nd brother of Z. Inflammation of bowels, sick 4 weeks, age 20. 1894. Z had just moved back to town, and was explaining to her family what had happened to her in the intervening ten years.
Z's father. Chest pains, 2 years after Z moved back. Inquest held. Young assistant carpenter later promoted to position of authority linked to Mary.
2nd rival banker. stomach ailment 8 years after construction of bank.
2nd brother-in-law of Z. 22 years after birth. Mine accident. Long delay, but he had recently moved in with Z and her husband.
Rival Banker 3. Indigestion, 18 years after construction of bank. Long, antagonistic history.
Mother of Z. Paralyzed 25 years after birth. By this time, she was living alone with the only surviving brother of Z.
3rd brother-in-law of Z. Left the US with his wife around time Z returned and the story broke. Returned to US alone in 1901. Died in mine accident 9 years later.
2nd husband. Infidelity and financial crimes 1918. Mary brought charges. Died, cause unknown, either during or immediately after trail.
Brother of 2nd husband. Financial crimes. Sick 1918. Sick/Poisoning 1921 (finally a poisoning analysis!). Gall ducts, liver, jaundice, erysipelas, delirious. Gastric hemorrhage and death 1922.
I get thirteen out of eighteen sicknesses/deaths within a short time after a Mary incident.
I await your feedback. Thanks, once again. Mike

 

[mnyquist]

Regarding the misdiagnoses. I believe cyanide was something rather new to the doctors at the time. As per my book:
"The large demand for cyanides for mining operations in the 1890s was met by George Thomas Beilby, who patented a method to produce hydrogen cyanide (prussic acid) by passing ammonia over glowing coal in 1892. This method was used until Hamilton Castner in 1894 developed a synthesis starting from coal, ammonia, and sodium yielding sodium cyanide, which reacts with acid to form gaseous HCN."

 

[mnyquist]

Regarding poisoning not affecting other family members:
Husband 1 was alone except for Mary and their son all three times.
Brother-in-law 1. Lived with his mother, so I don't know. Maybe drank something at work at the mines.
Banker 1. Died alone on a business trip.
2nd brother of Z. Don't know, but eventually all but one living at this location succumbed.
Z's father. Working with a lone young assistant all day on a barn.
Banker 2. At home, long duration sickness. Had wife + three children. So the event likely occurred not at home. Bank closed upon his death, accounts migrated to Mary.
Banker 3. Was on a business trip, staying in hotel. At a convention.
Mother of Z. Paralyzed. Lived alone except with sole surviving son.
2nd husband. Touring West Coast, staying alone in hotels.
Brother of 2nd husband. Travelled extensively.

 

[JeffM]

I have got to say that the more details that come out, the less plausible it becomes.

M’s husband has an affair and illegitimate child with Z. M kills husband but not Z or child. Instead, M kills husband of one of Z’s sisters using cyanide, which is misdiagnosed as consumption (a disease very well known in the 19th century). M waits another 22 years to arrange a fake mine accident to kill the husband of one of Z’s sisters.

Truly, **** hath no weirdness like a woman scorned: I’ll teach that woman not to mess with my husband by murdering her brothers-in-law. So the sample space consists of wronged wives. The probability that a wronged wife will murder her husband or the other woman within a few months is certainly not zero. The probability that a wronged wife will wait for 22 years to murder the brother-in-law of the other woman through a faked mining accident is effectively zero.

 

[mnyquist]

Hello again Jeff et al:

I thought this might be heading in the opposite direction. Rather than take into account the three sets of new supporting info I provided re: 1) the recent commercial development of cyanide making it more widespread, and perhaps as a result Doctor's unfamiliarity with it, and 2) the fact that the victims were alone without their families when they became ill, both of which answered very legitimate questions you had, you 3) ignored the several "incidents" which occurred immediately after incurring Mary's wrath, and concentrated on a single delayed revenge to refute the hypothesis. This seems to go against any probability analysis! That, and the fact that Z was spared.

I guess I need to go further into the story.

Mary's husband left for a victory tour to Europe soon after he impregnated Z. When he returned, he discovered the damage. Perhaps his wife had already heard rumors. Once again, he ignored his Congressional and business duties and jumped into problem-solving mode. Amongst his legions of miners, he identified a trusted manager who could help him in suggesting a candidate for a scapegoat husband. The strategy was for this person to marry Z and to move within the "empire" to a location far away from home, where they were to remain. The lieutenant, the husband's cousins, and father, were all moved to various distant locations. So was at least one other non-relative who knew about the affair. They did get married, and never registered the birth.

The deal was the illiterate and poor immigrant miner husband in return would be given positions of increasing responsibility, otherwise unavailable to him, in the organization. This in fact happened. For Z, it was likely explained that this was the only way he could guarantee the safety of her and her child.

Mary did her best to locate the couple, but could not. Husband E had covered their tracks. They were much too far away and no one was talking. Absolutely furious, she decided to track down relatives, which proved to be an easier task. Besides, these relatives were apparently on occasion spreading the story, which would have damaged the family reputation. It's a "you fractured my family: I'm going to show you what that's like" type of situation. Besides, she had to get revenge somewhere, and maybe this would draw Z out of the woodwork.

Yes, the 2nd person she killed was her husband. Soon after the affair, Mary began an open (revenge?) relationship with the son of a neighbor. In damage control mode, E did not run again in Congress, quit politics, concentrated on business in other regions, and, in an attempt to salvage his home life, attempted a move south. Mary already had other future plans, and that is why he experienced his 2nd and 3rd health issues during the move, at the meeting point and at the ending point. The end result was he was dead and wife and son were back where they started, and she back with her future husband #2. The southern mansion stood empty for years.

After ten years, Z, husband and daughter returned home. With E dead and Mary and son moved about one hour away, they thought they'd take a chance. But they were very diligent. The town printed a directory about every year, and as a result, every year, they moved. They also lived generally in one or another of the last houses on the edge of town. Two brothers and a father were dead, such deaths occurring around the time of her return.

The revenge against the 2nd husband and his brother was also swift.

So, I also have retaliations that took a long time. I had explained that there had been a lull, maybe due to things going well. But I also gave mitigating reasons: a move to Europe for several years, etc. The troubles with banker 3 and husband 2 predated Mary: perhaps she felt vengeance is sometimes best served cold, or perhaps the ill-feelings grew over time.

Or, throw out the long-awaited deaths, maybe they're wrong. I still have plenty that fit within a short time frame. Don't kill the probability because of one delayed reaction. I mean: you had asked for indications of a short time period between wrath and death, and I provided that, many times over.

Other comments? Cheers, Mike

 

[mnyquist]

I want to express my thanks to all those you responded to me in this thread. I gained many insights, and will eventually incorporate what I learned into a revised edition, resulting in a better product.
I started my investigation with Z's immediate family members. It spread out from there. No doubt, I overstepped and jumped to conclusions sometimes, as the findings became more bizarre. I guess I wanted to cover all the possibilities, but, in doing so, may have weakened my case.
This appears to certainly be the case of the first death, by consumption, as Jeff had indicated. This relative's connection was too remote, the cause of death too clear. The timing was also off: he died before Z returned and the story spread.
I will try to obtain further information. This is proving increasingly difficult.
It's kind of been like watching an episode of Forensic Files. Both husbands died in a very short time after "pissing off" Mary, one after being sick twice before.
And I still have the two brothers and father (not to mention the mother's illness) dying very soon after Z's reappearance in town after an absence of ten years.
I'll concentrate on that. But I also have much more information on the long, acrimonious relationships with the other bankers.
Once again, thanks. Couldn't work this into any probability equation, but it was a good, and interesting, trip!