Announcement

Collapse
No announcement yet.

The Margin For Error.

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • The Margin For Error.

    This point has arisen in various threads over the years so I thought that I’d gauge opinion without going into specifics and be accused of bias. So the question is this….

    When looking at all aspects of the case and we are trying to assess and interpret events should we allow a margin for error on all times? Not only estimated times but stated clock times (allowing for clock inaccuracy or poor synchronisation)? Or should we assume that all estimations are correct and that clocks and watches are all accurate and well synchronised?
    18
    Yes, we should always allow a margin for error.
    94.44%
    17
    No, we shouldn’t allow a margin for error.
    5.56%
    1
    Undecided.
    0%
    0
    Regards

    Sir Herlock Sholmes.

    “A house of delusions is cheap to build but draughty to live in.”

  • #2
    The problem with the margin of error concept as applied to subjective estimates of the time and the accuracy of clocks, is that margin of error is a statistical concept, but theorists will not use statistics to determine appropriate margins of error, but rather rely on their intuition to come up with a +/- number of minutes. This is pseudo-precision, but a worse problem is that these intuitions have a funny habit of being just enough to make their pet theories work. A supposed margin of error of 10 minutes is never used to explain away a 3-minute discrepancy - it will invariably be used to explain away a 10-minute discrepancy. This statistically unlikely situation is ignored - a 3-minute time error is quite likely when the margin of error is +/- 10 minutes, whereas a 10-minute error is highly improbable.
    Andrew's the man, who is not blamed for nothing

    Comment


    • #3
      A margin for error could certainly be used to keep someone’s theory in the game but I’d say that it’s more important not to dismiss a possibility on a matter of a few minutes. We know for a fact that these issues, clocks being fast or slow or poorly synchronised with other clocks, occur in the modern day therefore we know for certain that these would have happened then. But by assuming accuracy and perfect synchronisation we wouldn’t be using accurate methods. To be honest I don’t see a 10 minute discrepancy as highly improbable either when we consider our own lives. If anyone asked me what time I went to bed last night or the night before I would only be able to give an estimate which could easily be more than 10 minutes out in a house where I’m surrounded by clocks. We estimate time periods based on events so I could say “well I know that I did x at y o’clock and I went to bed around an hour later.” How many times have we all said that we’ve been waiting for something for so long only to be informed that we hadn’t actually been waiting anywhere near that long?
      Regards

      Sir Herlock Sholmes.

      “A house of delusions is cheap to build but draughty to live in.”

      Comment


      • #4
        Originally posted by NotBlamedForNothing View Post

        This statistically unlikely situation is ignored - a 3-minute time error is quite likely when the margin of error is +/- 10 minutes, whereas a 10-minute error is highly improbable.
        Which is absolutely right and those putting that case forward, move the times in a bit and begin to argue from their moved in times as though that is the information left to us, when in fact it's not at all.

        Comment


        • #5
          Not sure if this article is publicly available or not, but the below article covers the issue of clocks and reliability (in terms of the time they showed) in Britain. Basically, they were not reliable, nor in sync. And this continued into the early 1900s, and was becoming an increasing annoyance to people (apparently in 1908 there was quite a bit "action" to the Times, to the point the paper wrote an editorial on the unreliability of the clocks in London). Anyway, I've grabbed a bit from the article where it talks specifically about how the clocks were well known to be poorly syncronized, and generally told the wrong time. Unfortunately for us, the article doesn't mention anything we could use as a reasonable range. I think it fair to say it wasn't a couple minutes, or people would not complain so much. The one bit that mentions anything refers to a town in France, where noon could take up to half an hour to sound! But of course, that isn't London, so ...

          - Jeff


          From Gay, H. (2003). Clock Synchrony, Time Distribution and Electrical Timekeeping in Britain 1880-1925, Past & Present, 101, p 107-140.

          In 1908, Sir John A. Cockburn wrote a letter to the Times complaining about the clocks in London streets, none of which appeared to record the same, let alone the correct, time. As Cockburn put it, 'highly desirable as individualism is respects, it is out of place in horology'. And, he added, ‘a lying timekeeper is an abomination and should not be tolerated’. His letter hit a never and prompted further correspondence, and an editorial on the subject, suggesting that he was expressing a widely felt irritation, and that punctuality and knowing the time accurately were important to many people. Most public clocks in London were poorly regulated and church clocks were among the worst. In 1904, the Corporation of the City of London had recommended that all clocks ‘over-hanging the public way’ be synchronized but did not bother to comply by regulating its own clocks which remained unreliable.

          Similar complaints were heard elsewhere: ‘noon takes a quarter or even half an hour to sound’ was one made of town clocks in France.

          Comment


          • #6
            Originally posted by JeffHamm View Post
            Not sure if this article is publicly available or not, but the below article covers the issue of clocks and reliability (in terms of the time they showed) in Britain. Basically, they were not reliable, nor in sync. And this continued into the early 1900s, and was becoming an increasing annoyance to people (apparently in 1908 there was quite a bit "action" to the Times, to the point the paper wrote an editorial on the unreliability of the clocks in London). Anyway, I've grabbed a bit from the article where it talks specifically about how the clocks were well known to be poorly syncronized, and generally told the wrong time. Unfortunately for us, the article doesn't mention anything we could use as a reasonable range. I think it fair to say it wasn't a couple minutes, or people would not complain so much. The one bit that mentions anything refers to a town in France, where noon could take up to half an hour to sound! But of course, that isn't London, so ...

            - Jeff


            From Gay, H. (2003). Clock Synchrony, Time Distribution and Electrical Timekeeping in Britain 1880-1925, Past & Present, 101, p 107-140.

            In 1908, Sir John A. Cockburn wrote a letter to the Times complaining about the clocks in London streets, none of which appeared to record the same, let alone the correct, time. As Cockburn put it, 'highly desirable as individualism is respects, it is out of place in horology'. And, he added, ‘a lying timekeeper is an abomination and should not be tolerated’. His letter hit a never and prompted further correspondence, and an editorial on the subject, suggesting that he was expressing a widely felt irritation, and that punctuality and knowing the time accurately were important to many people. Most public clocks in London were poorly regulated and church clocks were among the worst. In 1904, the Corporation of the City of London had recommended that all clocks ‘over-hanging the public way’ be synchronized but did not bother to comply by regulating its own clocks which remained unreliable.

            Similar complaints were heard elsewhere: ‘noon takes a quarter or even half an hour to sound’ was one made of town clocks in France.

            Thanks for posting this Jeff.

            This really should put all doubt about the reliability and accuracy of clocks to bed once and for all (though I fear of course that it probably won’t ) Even 20 years after the murders clocks weren’t reliable and were the cause of complaint.

            That we should always allow for a margin for error when dealing with times should be considered the most basic of ‘rules’ when discussing any case from that era.
            Regards

            Sir Herlock Sholmes.

            “A house of delusions is cheap to build but draughty to live in.”

            Comment


            • #7
              Originally posted by Herlock Sholmes View Post

              Thanks for posting this Jeff.

              This really should put all doubt about the reliability and accuracy of clocks to bed once and for all (though I fear of course that it probably won’t ) Even 20 years after the murders clocks weren’t reliable and were the cause of complaint.

              That we should always allow for a margin for error when dealing with times should be considered the most basic of ‘rules’ when discussing any case from that era.
              Yah. It would be nice to have some more actual information about how far out clocks were, but I doubt that information exists in any systematic way. The letter writing to the Times, and maybe the editorial referred to, might have some references to how far out clocks could be, but I haven't tracked them down to look. I think a +-10 minute window seems a reasonable rule of thumb, so if two stated times are "out" by 10 minutes or less, and where both are getting their times from a clock, then that would be too close to be of concern. Wider than that might be a good reason to pause for thought.

              It gets more complicated if people are estimating the time, which they would be doing if they "saw a clock awhile ago, and figure it's been 15 minutes since then ..." as a whole new sort of error gets introduced (estimating durations of time), which people can be fantastically horrible at! In fact, the variability in that (between people) is so large it really is impossible to ever discount apparent discrepancies to be anything other than "guestimate errors". That's not to say that a large difference should be ignored, only that without further compelling evidence, that alone is no reason to draw a conclusion.

              - Jeff

              Comment


              • #8
                I’ve been reading the discussion board for the better part of this year and finally felt compelled to register. My knowledge of the details surrounding the Whitechapel murders is far more rudimentary than that of most Casebook members, but I have extensive knowledge of horology and large-format film photography (two subjects relevant to several discussions here).

                Over the last few years, I’ve read about a growing number of theories that heavily rely on relatively small discrepancies in time. I don’t have any favored theories on who committed the crimes – poor record management practices of the police and a world war destroyed most of the evidence, along with (in my opinion) any chance of conclusively finding the killer(s). That said, I urge great caution when considering any theories that heavily lean on single-digit time discrepancies.

                The clocks in Victorian London were not marine chronometers. Their escapements were more rudimentary, creating more friction than modern designs with more favorable materials and geometry. The teeth of the escapement wheel and ends of the pallet fork engaged in such a way that caused sliding between the parts, making them susceptible to variability if not regularly lubricated. Modern escapements, such as George Daniels’ coaxial escapement, utilize clever geometry to minimize sliding, reducing friction and lubrication requirements. Most movements of the time were also not well compensated for temperature – plain brass movements can slow up to 11 seconds per day per degree centigrade increase in temperature.

                Even modern mechanical watch movements of lower grades can be fast or slow by twenty seconds daily. Official accuracy standards such as Contrôle officiel suisse des Chronomètres (COSC) did not exist until the twentieth century.

                Comment


                • #9
                  Originally posted by Hyperfocal View Post
                  Their escapements were more rudimentary, creating more friction than modern designs with more favorable materials and geometry. The teeth of the escapement wheel and ends of the pallet fork engaged in such a way that caused sliding between the parts, making them susceptible to variability if not regularly lubricated. Modern escapements, such as George Daniels’ coaxial escapement, utilize clever geometry to minimize sliding, reducing friction and lubrication requirements. Most movements of the time were also not well compensated for temperature – plain brass movements can slow up to 11 seconds per day per degree centigrade increase in temperature.

                  Even modern mechanical watch movements of lower grades can be fast or slow by twenty seconds daily. Official accuracy standards such as Contrôle officiel suisse des Chronomètres (COSC) did not exist until the twentieth century.
                  Excellent first post and commentary. Welcome to the boards.

                  Let all Oz be agreed;
                  I need a better class of flying monkeys.

                  Comment


                  • #10
                    Originally posted by Hyperfocal View Post
                    I’ve been reading the discussion board for the better part of this year and finally felt compelled to register. My knowledge of the details surrounding the Whitechapel murders is far more rudimentary than that of most Casebook members, but I have extensive knowledge of horology and large-format film photography (two subjects relevant to several discussions here).

                    Over the last few years, I’ve read about a growing number of theories that heavily rely on relatively small discrepancies in time. I don’t have any favored theories on who committed the crimes – poor record management practices of the police and a world war destroyed most of the evidence, along with (in my opinion) any chance of conclusively finding the killer(s). That said, I urge great caution when considering any theories that heavily lean on single-digit time discrepancies.

                    The clocks in Victorian London were not marine chronometers. Their escapements were more rudimentary, creating more friction than modern designs with more favorable materials and geometry. The teeth of the escapement wheel and ends of the pallet fork engaged in such a way that caused sliding between the parts, making them susceptible to variability if not regularly lubricated. Modern escapements, such as George Daniels’ coaxial escapement, utilize clever geometry to minimize sliding, reducing friction and lubrication requirements. Most movements of the time were also not well compensated for temperature – plain brass movements can slow up to 11 seconds per day per degree centigrade increase in temperature.

                    Even modern mechanical watch movements of lower grades can be fast or slow by twenty seconds daily. Official accuracy standards such as Contrôle officiel suisse des Chronomètres (COSC) did not exist until the twentieth century.
                    Can I just Echo Ally's comment, a really good first post.
                    I have been arguing since for ever, that we should not use the times given by the witnesses as being little more than estimates.

                    Steve

                    Comment


                    • #11
                      I've been trying to get members to concentrate on a sequence of events as opposed to the time something is thought to have happened.

                      So long as we can establish 'B' happened before 'C', but after 'A', it really doesn't matter what time 'B' truly happened. We have this sequence between Hutchinson & Lewis, and between Diemschutz, Kozebrodski, Eagle, Spooner, and others, where the actual minutes involved are not as relevant when compared with the scene that was being played out.
                      Regards, Jon S.

                      Comment


                      • #12
                        Originally posted by Wickerman View Post
                        I've been trying to get members to concentrate on a sequence of events as opposed to the time something is thought to have happened.

                        So long as we can establish 'B' happened before 'C', but after 'A', it really doesn't matter what time 'B' truly happened. We have this sequence between Hutchinson & Lewis, and between Diemschutz, Kozebrodski, Eagle, Spooner, and others, where the actual minutes involved are not as relevant when compared with the scene that was being played out.
                        Exactly, I do the same with events in Bucks Row, relative timing, what happened in what order, and a rough estimate of how long after event A did event B occur.
                        The actual times events occur is secondary to how the events relate to each other.

                        Steve

                        Comment


                        • #13
                          I just want to endorse both Steve and Wickerman's posts. Stated times are generally estimates because most people, if they looked at a clock at all, did so some time prior to the event they are giving a time for (so they had to estimate what time it was at event X having looked at a clock previously). Off the top of my head, the only time stated which is based upon recording of a clock at the time of the stated event is Dr. Blackwell's arrival at the scene of the Stride murder, when he recorded the time by his own pocket watch. However, given none of the other witnesses will be referring to the time as shown on Dr. Blackwell's watch, then other stated times would be expected to be based upon clocks showing different times - but by how much we don't know. (Note, even Deimshutz's 1:00 was not recorded at the time, he later recalls the time he saw, he didn't write it down as he passed the clock).

                          But working through the testimonies, and ordering the events in sequence, results in a reasonable picture of the events. Some statements will stand out because we expect some statements to involve widely inaccurate estimates of the time, but in the end, the majority of statements should produce a coherent flow of events. We'll never get things exactly down to the minute, but we don't have to have that level of precision to understand the flow of the events. This is the approach I've adopted with the various simulations, and I believe it has resulted in some reasonable ideas.

                          - Jeff
                          Last edited by JeffHamm; 11-27-2023, 08:05 PM.

                          Comment


                          • #14
                            Originally posted by Elamarna View Post

                            Can I just Echo Ally's comment, a really good first post.
                            I have been arguing since for ever, that we should not use the times given by the witnesses as being little more than estimates.

                            Steve
                            What you've been arguing for amounts to a truism.

                            What we need to know is how approximate were various estimates of the time.

                            11:37:26 is 11:37 to the nearest minute.
                            11:37:26 is 11:35 to the nearest 5 minutes.
                            11:37:26 is 11:40 to the nearest 10 minutes.
                            11:37:26 is 11:30 to the nearest 15 minutes.
                            Andrew's the man, who is not blamed for nothing

                            Comment


                            • #15
                              Impossible to answer as you should be well aware. Two times by different people people can mean VERY different things
                              Such estimates of little real value.

                              However, If you watch the podcast on this site of the talk given at the 2022 East End Conference, you will see, the variations that were found by recording the discrepancies of modern day public clocks over 2021/22.

                              Steve
                              Last edited by Elamarna; 11-28-2023, 10:53 AM.

                              Comment

                              Working...
                              X