Chronometry: Has any of you come across a "chronometrical thermometer"?

Discussion in 'Chronometers' started by Luis Casillas, Dec 26, 2015.

  1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
  1. Luis Casillas

    Luis Casillas Registered User

    Oct 16, 2012
    570
    9
    0
    Software
    San Francisco Bay Area
    Country Flag:
    Region Flag:
    #1 Luis Casillas, Dec 26, 2015
    Last edited by a moderator: Dec 31, 2017
    In my reading about chronometers and observatory timekeeping I've now more than once come across mentions of a curious instrument that's alternatively called a chronometrical thermometer or thermometric chronometer.



    If you look at the Greenwich chronometer trials results from 1841/42 to 1915, you can see that the data tables record the weekly rates for the chronometrical thermometer along with the timepieces under trial. This tells us that the instrument was in use as early as 1841. Gardner's article, when it touches on the Greenwich trials, describes it (page 327):

    Leuschner's description (p. 177):

    Gardner says that Greenwich's instrument was compensated "the wrong way"; Leuschner says Lick's instrument was "uncompensated." I don't think it's clear whether the instruments were different or just the descriptions.

    Why would one want a chronometer that magnifies temperature fluctuations? Gardner tells us a bit about it (p. 327 again):

    Or in more mathematical terms, if we model temperature as a function of time, the chronometrical thermometer is indirectly measuring the integral of that function.

    So now, my question: has any of you seen anything like this out in the wild? So far it sounds like the thing to look for is a box chronometer where the balance is either:
    1. Obviously not compensated;
    2. Has a "compensation" whose design appears to be backwards.
     
  2. Luis Casillas

    Luis Casillas Registered User

    Oct 16, 2012
    570
    9
    0
    Software
    San Francisco Bay Area
    Country Flag:
    Region Flag:
    #2 Luis Casillas, Dec 26, 2015
    Last edited: Dec 26, 2015
    I found two more references, these from the United States Coast Survey in the latter 1850s:



    The first reference describes their instrument in a bit of detail (page 183):

    The second reference mentions that they used the exact same instrument as in the first.
     
  3. Luis Casillas

    Luis Casillas Registered User

    Oct 16, 2012
    570
    9
    0
    Software
    San Francisco Bay Area
    Country Flag:
    Region Flag:
    One more:

    • Belville, John Henry. 1850. A Manual of the Thermometer, Containing its History and Use as a Meteorological Instrument. London: Richard and John Edward Taylor.
    On page 9:

    So it sounds like there may have been two methods of construction of balance wheels for these instruments—plain brass (as in the US Coast Survey instrument) vs. "backwards" bimetallic (as described here by Belville, who was an assistant at Greenwich).
     
  4. MartyR

    MartyR Moderator
    NAWCC Member

    Dec 16, 2008
    10,927
    250
    63
    UK
    Country Flag:
    If I hadn't seen all those learned articles you guys have found, I would have said this was a hoax :D

    I don't believe there are any scientific research papers which would enable a user of this device to measure accurately the average temperature over time from the time gain/loss. I don't believe that because none of the papers quoted seems to to refer to such a research paper, and if the scientific research had been done, they would have referred to it. That forces me to the conclusion that all a user could determine was that the more the time lost or gained by the device, the nearer or further the average temperature over time would have been to or from the maximum. That supposition is supported by the comments in one or two of the articles.

    If that is so, then why on earth would one need to use a chronometer at all? A standard non-chronometric watch would produce exactly the same results.
     
  5. Ralph

    Ralph Registered User
    NAWCC Member Sponsor

    Jan 22, 2002
    4,662
    61
    48
    Country Flag:
    #5 Ralph, Dec 26, 2015
    Last edited: Dec 26, 2015
  6. Tom McIntyre

    Tom McIntyre Technical Admin
    Staff Member NAWCC Star Fellow NAWCC Ruby Member Sponsor

    Aug 24, 2000
    81,776
    1,292
    176
    Male
    retired SW dev
    Boston
    Country Flag:
    Region Flag:
    Marty, the only reason to build it like a chronometer would be to eliminate the variation from other factors such as isochronism primarily. The barometric error would also contribute but would be much smaller.

    I do not think anyone would feel the need to do research on the device except possibly to resove other sources of deviation in rate.

    The fact that it is an integral thermometer is clear from the design. Either the plain balance or the reversed compensation balance would work, but the second gives a larger value for time change as a function of the integral of the temperature over time.

    The recording thermometer could also be used with a planimeter (area integrator) for the same purpose.
     
  7. MartyR

    MartyR Moderator
    NAWCC Member

    Dec 16, 2008
    10,927
    250
    63
    UK
    Country Flag:
    What I'm saying is that if there was a mathematical formula or a graph from which one could calculate the average temperature over time from the deviation of the device's time reading from that of the regulator, then I could see a good reason for using a chronometer. But if all that was possible was to say that the mean temperature over time was somewhere above or below the mean, then I can't see a good reason. The extracts I have read in this thread suggest that the latter was the case, although I accept that reading the full documents might show something different.
     
  8. Tom McIntyre

    Tom McIntyre Technical Admin
    Staff Member NAWCC Star Fellow NAWCC Ruby Member Sponsor

    Aug 24, 2000
    81,776
    1,292
    176
    Male
    retired SW dev
    Boston
    Country Flag:
    Region Flag:
    The temperature error of a compensated chronometer with or without middle temperature correction is a function of the temperature over the period of study. That sum of positive and negative errors is what is of interest to the designer. The thermometric chronometer makes it possible to observe that phenomenon directly. i.e. the cumulative error of the compensated chronometer was xxx and the cumulative error of the thermometric chronometer was yyy. The percentage of the error corrected is a constant times yyy-xxx.

    The other approach to measuring the effectiveness of compensation is to use an oven and a refrigerator but, while that can be very precise, it does not reflect real world use.
     
  9. DeweyC

    DeweyC Registered User
    NAWCC Member

    Feb 5, 2007
    1,459
    371
    83
    Watchmaker
    Baltimore
    Country Flag:
    Regarding Richard Schmidt: He is a customer and in fact I restored that instrument. You can clearly see it is a bimetallic balance; it is difficult to miss the blued steel inner lamination. It is not a plain brass balance, it is a typical compensated chronometer balance.

    From the description provided by Luis, it looks like the whole idea was to develop an instrument that indicated a mean temperature for every instantaneous moment of a given period. As he said, an integral. Seems like a long way to go around the barn, and may be an interesting dead end.

    It was known long before 1840 that a simple brass annulus coupled to a steel spring was very sensitive to temperature. The whole point of making a bimetallic balance and coupling it to a hardened steel spring was to compensate for the effect of temperature of the steel spring by having the balance assume a smaller moment of inertia in high temperature (spring softer) and a larger moment of inertia in lower temperatures (spring harder). The affixes were intended to adjust for the fact that the spring changes were linear while the balance changes were not, resulting in a middle temp error.

    This is discussed in an article I wrote for the Bulletin in the early 1990s in which I showed examples of Hartnup, Guillaume, Poole, both Kullbergs, etc and described how they helped correct the MTE.

    Smart people tried a lot things in the 19th century; some actually panned out. But like a dear friend was fond of saying "You gotta have ideas before you have good ideas".
     
  10. Luis Casillas

    Luis Casillas Registered User

    Oct 16, 2012
    570
    9
    0
    Software
    San Francisco Bay Area
    Country Flag:
    Region Flag:
    Thanks everybody for your comments. And Ralph, that's a great link!

    I do not get the impression so far that most of the users of this instrument were trying to measure the integral of temperature directly as its own quantity, but more like trying to obtain some quantity that varied as a linear-ish function of it, to which the rates of conventional chronometers could be compared. For example Gardner says (p. 328):

    Since the Greenwich trials results are publicly available, this sounds like a question that could be examined more carefully with some statistics; how do the rates of the various chronometers correlate with those of the chronometrical thermometer that accompanied them through the tests? I'm hoping to teach myself some stats, so I might have a shot at this when I get some time.
     
  11. Luis Casillas

    Luis Casillas Registered User

    Oct 16, 2012
    570
    9
    0
    Software
    San Francisco Bay Area
    Country Flag:
    Region Flag:
    One more citation:

    Airy, G.B. 1842. "Report of the Astronomer Royal to the Board of Visitors." Greenwich Observations in Astronomy, Magnetism and Meteorology made at the Royal Observatory, Series 2, vol. 3, pp.D1-D7.

    Here we have Airy crediting Molyneux for the instrument, and mentioning that it's bimetallic:

     
  12. Luis Casillas

    Luis Casillas Registered User

    Oct 16, 2012
    570
    9
    0
    Software
    San Francisco Bay Area
    Country Flag:
    Region Flag:
    Well, here's a puzzle. As is well known, a regular brass/steel bimetallic balance wheel is subject to middle temperature error, meaning that if the chronometer is regulated to keep correct time at (say) 45 and 85 degrees, it will be fast between those temperatures, and slow outside of that range.

    Molyneux's chronometrical thermometer's balance, as per Airy's description cited above, has "the laminae of its compensation bar reversed." So unless it has some for of auxiliary compensation that hasn't been described, it should be subject to reverse middle temperature error. It should run "too slow" at middle temperatures and "too fast" at extreme ones.

    So now what do we make of this scatterplot? Each point is a pair of sums of weekly rates—x axis is the chronometrical thermometer, y axis is the chronometer:

    Molyneux 2166 (1842).png

    Clearly at least one of the instruments here is affected by a middle temperature error, but how do we untangle the thermometer's contribution from the chronometer's? (This perhaps supports Marty's incredulity about the usefulness of this instrument.)
     
  13. Tom McIntyre

    Tom McIntyre Technical Admin
    Staff Member NAWCC Star Fellow NAWCC Ruby Member Sponsor

    Aug 24, 2000
    81,776
    1,292
    176
    Male
    retired SW dev
    Boston
    Country Flag:
    Region Flag:
    Who made the plot?

    It looks to me like the temperature effect on the Molyneux was set to be compensated near the plus and minus 750 "chronodegrees" of the thermometric chronometer. I do not know what those represent in terms of temperature but presumably + is hotter and - is colder. The Molyneux shows a middle temperature error as might be expected for a standard bi-metalic balance and steel spring. The great bulk of the graph is outside the normal operating range of the chronometer and it loses substantially at both extremes.
     
  14. Luis Casillas

    Luis Casillas Registered User

    Oct 16, 2012
    570
    9
    0
    Software
    San Francisco Bay Area
    Country Flag:
    Region Flag:
    #14 Luis Casillas, Jan 19, 2016
    Last edited: Jan 19, 2016
    I did. The data is straight from an 1842 Greenwich trial results table.

    I was getting mixed up and misinterpreting the effect of a middle temperature error on the chronometrical thermometer. The instrument's response to temperature is not linear, but the "chronodegrees" value is in all certainty a monotonic function of the integral of temperature nevertheless—that's the part that I got mixed up on. What the non-linearity of the thermometer would do is make it harder to interpret the "chronodegrees" as temperatures.

    So yeah, it's as you say:

    Which I think then suggests that the chronometrical thermometer is fairly effective at detecting middle temperature errors. I'll have to compare it to the min/max recorded temperatures though.
     

Share This Page