Stuart Kind, Geographic Profiler.

[The Rookie Detective]

Jeff, can I add a spanner to the works by asking how the Ripper map would change IF the Torso murders were included?

I know it's made trickier in that only Elizabeth Jackson has ever been identified from all the torso victims, but perhaps the locations where the killer chose to dump each torso is arguably more applicable for the implementation of Geo-profiling?

Just to add that Jackson was seen by multiple independent witnesses talking with a man who looked like a "Navvy" (a civil engineer who worked on the construction of the railways/waterways) ...just a few hours before she was murdered and dismembered.

I believe this man was the torso killer.

I also believe that the Torso and Ripper killings were linked, but that the Ripper wasn't necessarily the same man as the torso killer.

And so my question is; how would adding the Torso locations impact the Geo-profiling map that you've created for the Ripper murders?


RD

[JeffHamm]

Jeff,
this really seems to be your wheelhouse, so I wonder if you would have any idea how much the YR geo-profile would have changed had the Sheffield incident gone the other way?

Do you just mean what if he wasn't arrested when he was?

It's hard to say how, or if, the profile would change much. There's a fair amount of data points in there already, which usually results in a fairly stable spatial output - partly because offenders tend to just keep doing what they're doing and going to the same places they're already going - and if they venture into a new area, it usually will just add more information, tightening up the results (meaning even if he killed there, he would have killed many more in his usual areas).

So, had he not been arrested, he would have continued killing, that's for sure. And if we say he murdered the woman he was with when arrested, which I think is what you're getting at, the spatial output wouldn't alter much. I've roughly tried to estimate where Sheffield is, and I've added it just outside the borders near the 100 on the x-axis. Probably should be a touch further south, but it is close enough for our purposes.

And then re-ran things, and we get this (again, just using my own routines), and as you can see, his two residences are still in Zone 1. Really, the only thing that would have happened is that the estimated crime range size grows a bit as Sheffield would have expanded the smallest circle that encloses all of the offenses a bit, and as a result, the area in each zone will increase. But overall, where the high interest area gets located isn't going to suddenly jump to Sheffield, or even shift all that much. What is a bigger problem is when an offender moves to a new location. Some return to the areas they were used to, but as they gradually become familiar with their new surroundings, and that means the anchor points that are the underlying structure of their spatial behaviour changes, and that starts to change the "shape" of the output a lot more. Changing job locations can have the same influence.

One of the ideas I would like to try would be to find a fairly lengthy series where the offender does just that, makes a big move in the middle of it (let's say for 20 offenses, moving after offense 10). Then, produce a profile from the first 5, let's say, then drop the earliest and add offense 6, do a profile, and keep dropping the earliest and adding the next. What one would hope to see is a fairly regular and stable pattern, that starts to change once the offender has moved, and then settles into a new stable pattern. Sutcliff moved, but not sufficiently far for me to think it it would make a noticeable difference. Mind you, that doesn't mean it wouldn't work, as I've not tried it. Certainly there would be quite a shift when the Manchester locations get added though, as the smaller number of offenses needed for what I'm thinking here would be less stable (and, in the end, Manchester was an anchor point for him as his job took him there quite often, so it was part of his normal routines, so introducing a new anchor point should make a difference. Hmmm, maybe I've just convinced myself that it would be worth doing with this after all! ha ha).

[The Rookie Detective]

The fundamental issue with applying Geo-profiling to Sutcliffe, is the fact that there were numerous other attacks that he carried out, that have not been officially attributed to him.
On that basis; the data is incomplete and renders it ineffective, because unless ALL of his attacks are included, we can't gain a fully rounded accurate picture.

Having pieces missing from a Jigsaw; will impact our ability to see the full picture.


RD

[Fleetwood Mac]

The fundamental issue with applying Geo-profiling to Sutcliffe, is the fact that there were numerous other attacks that he carried out, that have not been officially attributed to him.
On that basis; the data is incomplete and renders it ineffective, because unless ALL of his attacks are included, we can't gain a fully rounded accurate picture.

Having pieces missing from a Jigsaw; will impact our ability to see the full picture.


RD

This isn't supported by geographic profilers.

The reason being that there are instincts and factors which drive human behaviour, and they will follow a pattern. Driven by concepts such as rational choice, comfort zones and so on.

There is a real life example of one such geographic profiler who did not know all of the attacks but did deduce where Sutcliffe lived. The report, including his logic, is posted on this thread and that will give you a better idea of the methods involved and why they do not need to know all of the attacks.

What's probably more noteworthy, is that the concept of a 'buffer zone' is hotly disputed among people in that profession. I didn't realise that until reading a few articles this morning. Empirical studies have been undertaken on this, and it seems there is even disagreement among people in that field as to whether or not a 'buffer zone' exists when it comes to serious crime.

[JeffHamm]

The fundamental issue with applying Geo-profiling to Sutcliffe, is the fact that there were numerous other attacks that he carried out, that have not been officially attributed to him.
On that basis; the data is incomplete and renders it ineffective, because unless ALL of his attacks are included, we can't gain a fully rounded accurate picture.

Having pieces missing from a Jigsaw; will impact our ability to see the full picture.


RD

Obviously having all the information is the ideal, but the thing about patterns is that a lot of the information becomes redundant, so actually having only a subset is fine. Also, well established patterns are robust and stable from the odd deviation from the pattern (as is shown in my previous post).

If you look at the analysis of the Sutcliff case above, you can see it places his residence in the highest priority zone, so that looks pretty effective to me.

[JeffHamm]

Jeff, can I add a spanner to the works by asking how the Ripper map would change IF the Torso murders were included?

I know it's made trickier in that only Elizabeth Jackson has ever been identified from all the torso victims, but perhaps the locations where the killer chose to dump each torso is arguably more applicable for the implementation of Geo-profiling?

Just to add that Jackson was seen by multiple independent witnesses talking with a man who looked like a "Navvy" (a civil engineer who worked on the construction of the railways/waterways) ...just a few hours before she was murdered and dismembered.

I believe this man was the torso killer.

I also believe that the Torso and Ripper killings were linked, but that the Ripper wasn't necessarily the same man as the torso killer.

And so my question is; how would adding the Torso locations impact the Geo-profiling map that you've created for the Ripper murders?


RD

Hi RD,

I don't have a map with all the torso locations. As I recall, it is likely some were thrown in from a bridge and they drifted to where they were found. The important location would be the bridge, as that location was chosen by the killer.

If you have a map with the ripper killings and the torso locations, I could probably do one though.

- Jeff

[GBinOz]

Anyway, as for Martha Tabram though, I think George Yard Buildings to the south is where Martha was killed. Curiously, it appears her last known address was 19 George Street, Spitalfields (known as Satchell's Lodging House), and while I have located the street on the map yet, Spitalfields would be to the North of Flower and Dean. I haven't found Gun Street either, but it doesn't appear to be in the immediate area, so I've not looked too far from there.

Oh, I didn't realise we knew where Richardson lived! That's a new one on me.

Hi Jeff,

My mistake, Tabram did live at Satchell's Lodging House, which was between Flower and Dean and Thrawl St. Gun St is west of the end of Dorset St, next street over from Crispin St where it is claimed Thompson was living at the time. McKenzie lived on the eastern side of Gun St, not far north of the intersection with Artillery St. Richardson lived at 2 John St (now Wilkes) on the western side a few doors from the intersection with Hanbury St.

Best regards, George

[The Rookie Detective]

Hi RD,

I don't have a map with all the torso locations. As I recall, it is likely some were thrown in from a bridge and they drifted to where they were found. The important location would be the bridge, as that location was chosen by the killer.

If you have a map with the ripper killings and the torso locations, I could probably do one though.

- Jeff

Yes, I agree that some of the body parts were thrown into the water.

However, the Pinchin Street Torso and The Whitehall torso were placed under a railway arch and under the foundations for the New Scotland Yard building respectively.

Elizabeth Jackson is trickier because she was dismembered and dumped in the water, and so the Thames itself, would act as the Southern Boundary so to speak.

The 2 locations NOT involving water are significant IMO, and it would be interesting to see how Pinchin Street and Whitehall Mystery impact the Geo-Profiling for the Ripper.


There's an age-old argument that a killer doesn't change their M.O...but there are many cases of serial killers who are non-ritualistic in their application, and have altered their MO and adapted accordingly; the thrill of the kill being the primary driving force.
Killers may have a main go-to way of killing their victim, but serial killers are often adaptable and experiment with different ways to achieve the same outcome.

If we did consider the Torso Killer and the Ripper as being the same killer, it would broaden the scope.

I am on the fence, but I do find the timing, proximity, use of the knife, and choice of victims between the two cases, too similar to ignore.


RD

[GBinOz]

My understanding is that the Flower & Dean/Dorset street area had a high concentration of "unfortunates" relative to other, even fairly nearby, areas. Not sure where I get that from though? If unfortunates were equally distributed over the whole area, then the odds of getting 5 from the same 1/3 area, would be 1/3^4 (4 not 5, as the first victim could come from anywhere, after that the other 4 have to come from the same "1/3" area as the first), which is about 1.2%. But if "unfortunates" clustered in that area (woman tended to group together in those doss houses), then the odds could be higher. Did Tabram, McKenzie, Coles, or some of the other victims, have connections to that area as well? I can't recall. If so, it may be that it isn't so unlikely after all.

Hi Jeff,

These are quotes from the same source:
THE NUMBER OF LODGING HOUSES
His letter also revealed just how widespread the Common Lodging Houses were throughout the relatively small area that was bounded by Baker's Row to the east, Middlesex Street to the west and Whitechapel Road to the south:-
...There are no less than 146 registered lodging-houses, with a number of beds exceeding 6,000. Of these 1,150 are in Flower and Dean-street alone, and nearly 700 in Dorset-street. Some of the houses contain as few as four beds, whilst others have as many as 350. At a few of these men only are received, and at others women only, but in the majority there are what are known as "double-doss beds" ... there is little room to doubt the truth of the assertion that when these double beds are let no questions are asked, and the door is opened for the most frightful immorality..."

THE NUMBER OF PROSTITUTES REVEALED
Based on the observations of the H division constables, whose beats took in the district to the west and east of Commercial Street, the police set the number of Common Lodging Houses at 233, the number of residents at 8,530 and the number of brothels at 62.

The police reply also stated that "we have no means of ascertaining what women are prostitutes and who are not, but there is an impression that there are about 1200 prostitutes, mostly of a very low condition."

It is not altogether clear whether the observations of the H division beat constables took in the whole of H -Division, which is shown on the map below:



Best regards, George

[GBinOz]

If we did consider the Torso Killer and the Ripper as being the same killer, it would broaden the scope.

I am on the fence, but I do find the timing, proximity, use of the knife, and choice of victims between the two cases, too similar to ignore.

RD

Hi RD,

It appears that we have the alternatives of one serial killer with multiple victims, or multiple serial killers operating at the same time in the same area. If the latter, we have the problem of allocation of victims to perpetrators. We'll probably never know which is correct.

Cheers, George

[The Rookie Detective]

Hi RD,

It appears that we have the alternatives of one serial killer with multiple victims, or multiple serial killers operating at the same time in the same area. If the latter, we have the problem of allocation of victims to perpetrators. We'll probably never know which is correct.

Cheers, George

I agree George


I think the likelihood of knowing whether the cases are connected, is something that will never be known.

On balance, I do believe that there was a link between the 2 cases.


I do have a hypothesis based on that idea that I'd like to share, in order to get some feedback, but it would require me to start another new thread; so that I don't unfairly go off-topic on this thread.
There are countless hypotheses and theories of course, but I strive to be as original as I can be; all the while recognizing and respecting that I am standing on the shoulders of giants.


RD

[A P Tomlinson]

Do you just mean what if he wasn't arrested when he was?

It's hard to say how, or if, the profile would change much. There's a fair amount of data points in there already, which usually results in a fairly stable spatial output - partly because offenders tend to just keep doing what they're doing and going to the same places they're already going - and if they venture into a new area, it usually will just add more information, tightening up the results (meaning even if he killed there, he would have killed many more in his usual areas).

So, had he not been arrested, he would have continued killing, that's for sure. And if we say he murdered the woman he was with when arrested, which I think is what you're getting at, the spatial output wouldn't alter much. I've roughly tried to estimate where Sheffield is, and I've added it just outside the borders near the 100 on the x-axis. Probably should be a touch further south, but it is close enough for our purposes.

And then re-ran things, and we get this (again, just using my own routines), and as you can see, his two residences are still in Zone 1. Really, the only thing that would have happened is that the estimated crime range size grows a bit as Sheffield would have expanded the smallest circle that encloses all of the offenses a bit, and as a result, the area in each zone will increase. But overall, where the high interest area gets located isn't going to suddenly jump to Sheffield, or even shift all that much. What is a bigger problem is when an offender moves to a new location. Some return to the areas they were used to, but as they gradually become familiar with their new surroundings, and that means the anchor points that are the underlying structure of their spatial behaviour changes, and that starts to change the "shape" of the output a lot more. Changing job locations can have the same influence.

One of the ideas I would like to try would be to find a fairly lengthy series where the offender does just that, makes a big move in the middle of it (let's say for 20 offenses, moving after offense 10). Then, produce a profile from the first 5, let's say, then drop the earliest and add offense 6, do a profile, and keep dropping the earliest and adding the next. What one would hope to see is a fairly regular and stable pattern, that starts to change once the offender has moved, and then settles into a new stable pattern. Sutcliff moved, but not sufficiently far for me to think it it would make a noticeable difference. Mind you, that doesn't mean it wouldn't work, as I've not tried it. Certainly there would be quite a shift when the Manchester locations get added though, as the smaller number of offenses needed for what I'm thinking here would be less stable (and, in the end, Manchester was an anchor point for him as his job took him there quite often, so it was part of his normal routines, so introducing a new anchor point should make a difference. Hmmm, maybe I've just convinced myself that it would be worth doing with this after all! ha ha).

Sorry Jeff, what I meant was, that he had the girl in the car, he had the tools in reach. If he hadn't been arrested he would probably have attacked her, likely killed her.
How would an attack that far south affect the mapping of the model?

(accidentallly deleted this )
The area doesn't seem to ahve moved that much which surprises me. (Not in a "I don;t believe it" way, but more in a "I thought it would have done more" way)

Back in the early 90s when I was a "mature" student at Wakefield College, I did a lot of part time work at the local Arts Centre as Technical Crew, and we hosted a series of crime seminars. One was a pre release screening of "Let Him Have It" where I got to meet Tom Courtenay and a young Chris Ecclestone as they were part of a live Q&A with Iris Bentley as they were trying to get a posthumous pardon for Derek. One of the others was related to the Yorkshire Ripper Investigation, and one of the few things that I remember from it was someone who worked on the investigation saying that if he had got away with the murder in Sheffield, then it would have massively impacted the search area and potentially become a "pissing contest" between West and South Yorkshire Police Forces.
So I'd always thought that the way geo profiling works, adding a factor that would have been in the opposite direction to the focus of the search would have brought the foacl area of the profile more in line with that drag effect. If that makes sense?

[Scott Nelson]

Either way, whether you believe the number of victims was five or six, or you believe all of the attacks in that period were attributable to the WM, or somewhere in between; Kind's methods would not lead to Flower and Dean Street and the surrounding streets.

You would end up somewhere just off the Whitechapel Road in an area encompassing the London Hospital and the eastern end of Hanbury Street.

Oh, OK then. How about 254 Whitechapel Road? Could that be close?

[JeffHamm]

Sorry Jeff, what I meant was, that he had the girl in the car, he had the tools in reach. If he hadn't been arrested he would probably have attacked her, likely killed her.
How would an attack that far south affect the mapping of the model?

(accidentallly deleted this )
The area doesn't seem to ahve moved that much which surprises me. (Not in a "I don;t believe it" way, but more in a "I thought it would have done more" way)

Back in the early 90s when I was a "mature" student at Wakefield College, I did a lot of part time work at the local Arts Centre as Technical Crew, and we hosted a series of crime seminars. One was a pre release screening of "Let Him Have It" where I got to meet Tom Courtenay and a young Chris Ecclestone as they were part of a live Q&A with Iris Bentley as they were trying to get a posthumous pardon for Derek. One of the others was related to the Yorkshire Ripper Investigation, and one of the few things that I remember from it was someone who worked on the investigation saying that if he had got away with the murder in Sheffield, then it would have massively impacted the search area and potentially become a "pissing contest" between West and South Yorkshire Police Forces.
So I'd always thought that the way geo profiling works, adding a factor that would have been in the opposite direction to the focus of the search would have brought the foacl area of the profile more in line with that drag effect. If that makes sense?

In the end we're dealing with very complex behaviours. Human behaviour and choices are not simple things, the way physics and chemistry are. The physical sciences study one participant, the physical universe, while each and every person is a universe unto themselves. If physics, each time it took a measurement of, for example, gravity, had to take a measurement from a different universe, where the gravitational constant could vary from one to the other (and, as the universe progressed, that too could change in various ways, and sometimes it could for some reason decide to override that value, etc), then our understanding of physics would likewise be less exact.

Also, do bear in mind, the above zones is the 75% area, meaning, 75% of the time this is where you'll find the offender, but that means 25%, or one in four, they are outside that zone! From an investigative view, that still creates a thinning of resources. Sure, hindsight tells us that in this case it worked, but an ongoing investigation doesn't get to see where the offender's residence (in this case) is actually located, and all they know is that there's a good chance they are somewhere in there. And because of the size of Sutcliff's criminal range, that area is still pretty large. Add Sheffield in, and there then becomes another police force involved, and the whole possible range expands.

Again, spatial analysis is about creating an organized search pattern that helps to optimise the search, it doesn't "do the search" though. And as with anything, there will be times when an offender doesn't conform to the norm, that's the thing about probabilities, you expect some cases where it doesn't appear to work. When faced with just the offense pattern, and a suggested search map like the above, how do you know if you're dealing with one of the 3 out of 4 cases where that search will help or do you have the 1 in 4 where they're somewhere outside of that?

As I say, it's not magic, it's just maths. it is useful information to keep in mind, but it doesn't do the thinking, and if an investigation is leading the police to a suspect that isn't in the search zone, then ignore the spatial analysis and follow the actual evidence. These are suggestions as to where to look for evidence, but they are not evidence in and of themselves anymore than being the spouse of a murder victim is evidence you are the offender! It just means you are in a group where a high percentage of such offenders can be found. That's quite a different thing, but for the investigator, that means they will want to check you out. And if they follow that people search pattern regularly, over the course of their investigations they will find offenders faster than if they just search the population randomly.

Spatial analysis is often presented in the news, or entertainment, as if it "solves" cases. It doesn't, all it does is provide a source of information to consider. it is empirically demonstrated to provide useful information, but in the end it is simply a prioritized search pattern, not the results of the search itself.

Anyway, sorry to go on so much, but this is one of those things where the public perception is often widely off the mark due to how "profiling" is presented in the press and entertainment, neither of which are accurate descriptions. It doesn't help when there are some researchers whose presentations are a bit over zealous, often due to having a financial interest in selling software. As a result, people either think the goal is to "solve the case", or they view it as a complete waste of time; neither of which is justified. It's not a solution, true, but that doesn't mean it's a waste of time either.

- Jeff

[JeffHamm]

Hi Jeff,

My mistake, Tabram did live at Satchell's Lodging House, which was between Flower and Dean and Thrawl St. Gun St is west of the end of Dorset St, next street over from Crispin St where it is claimed Thompson was living at the time. McKenzie lived on the eastern side of Gun St, not far north of the intersection with Artillery St. Richardson lived at 2 John St (now Wilkes) on the western side a few doors from the intersection with Hanbury St.

Best regards, George

Ah, on the map the street running N-S connecting Flower and Dean and Thrawl is shown as Lolesworth Street, so appears to have undergone a name change (the map is for 1892 - 1914, and some street names were changed; but it gives the best detail of the layout. Just a bit of a pain at times; Dorset Street, for example, now shows as Duval on this map). Anyway, that would place Martha in the same area. Found Gun Street (oddly, I had scanned around Artillary Lane, but missed it as it was a quick search). Also, I checked Coles' victim page, and she and Sadler stayed at 8 White's Row, which is one street down from Dorset, so I've put a marker on White's Row as well.

I've quickly mapped out the locations, with the blue dots showing residences (Richardson at the top; McKenzie on the left, and a general marker for Flower and Dean for Eddowes and Stride, and another for where I think you're referring to with regards to Martha Tabram's lodging house running between Flower and Dean and Thrawl Street, and Thrawl for Nichols; a few I've just placed in the middle of the street as general indicators as I don't have to hand exact locations). The red dots show the Chapman and Kelly crime scenes, with Kelly's location indicating Dorset Street, and the blue marker indicates Dorset Street, where Annie lived, but I'm not positive as to the exact location of Crossinghams, and the one below is White's Row, where Coles had stayed. The yellow marker is the GSG.

With regards to the C5, then we're looking at an area covering from Dorset Street over to Flower and Dean/Thrawl. Tabram falls in that area as well, as does Coles. McKenzie's location on Gun Street is about the same distance to the west end of Dorset Street/White's Row as the east ends are to Flower and Dean/Thrawl, so I'm not really sure one could say she was from a "different area" really.

In other words, I'm not so sure that one could say the fact the C5 lived "close" to each other is really improbable, given that one could say the same about McKenzie, Coles, and Tabram as well (they span a very similar area). I suppose if one includes them as potential JtR victims, one might see this as a further indication, but personally, I think it just shows that the people living in this area were people who, due to their circumstances, were at high risk.

- Jeff