PURPOSE: The purpose is to determine the accuracy of two mechanical watches: 1) Glycine Airman 2000 Automatic and 2) Record-Ball pocketwatch Caliber 435B. This is a collection of data gathered by comparison of each watch's time against a control timekeeper.
EQUIPMENT
| 1 | Glycine Airman 2000 Automatic wristwatch. Serial number 3764? | |
| 2 | Record-Ball pocketwatch, Caliber 435B, Serial number 0907 | |
| 3 | Control timekeeper: Computer clock synchronised via Network Time Protocol to GPS atomic clock |
DATA COLLECTION METHOD: A data point is gathered by holding the watch near the control timekeeper and noting the time displayed on each at a given instant. This is done every so often as time permits.
GRAPHING METHOD: The data points are recorded on computer. They are automatically gathered, graphed and presented in this forum via the links below using the following toolchain: Data --> converted to seconds --> graphed using GnuPlot.
Data| Airman 2000 | Data (text) | Graph (JPEG) | Data/Graph (PDF) |
| Record-Ball | Data (text) | Graph (JPEG) | Data/Graph (PDF) |
| Both watches | Data (text) | Graph (JPEG) | Data/Graph (PDF) |
The Record-Ball pocketwatch was my father's watch, used when he worked for Canadian National Railroad in Winnipeg, Manitoba. Read this email from an pocketwatch enthusiast who kindly responded to my request for information about the watch. Also have a look at this website regarding pocketwatches. If you scroll down the page, you'll see the movement of a watch very similar to mine. Here's how many Record-Ball 435 Caliber pocketwatches were made.
I bought the Airman 2000 in 2001 from The Falcon Watch Co. for everyday use. Unfortunately, at the time of this writing, it has not been accurately adjusted (as you can see from the graphs above). I have plans to send it out for adjustment, but we'll see how that goes.
So why all this crap? Well, it all started out when I noticed that the watches were running fast. So I started taking data points every once in awhile. After I had a few data points, I decided it might be worth it to look for trends, so I wrote a little perl script to read the data file and convert the values to seconds, so that I could graph them with GnuPlot. Then I built a set of shellscripts that would automatically generate the graphs, convert them to JPEG files and pack all the data into PDFs. It's not completely automatic. I have to run one script after every time I enter a new data point.
What tools did I use? Let's see, there's:
Example datafile: Record-Ball Pocketwatch data (OK, not really a tool per se) The perl script I use to convert my handata points into graphable data GnuPlot, an excellent graphing application Example GnuPlot command file: Tells GnuPlot how to produce this particular graph (Not Really a Tool) Example script that creates the graph, makes the PDF and JPEG files MPage, maintained by Mesa, which I use to combine the graphs and data together into a PDF, along with... ps2pdf, part of GhostScript, which turns PostScript files into PDFs The master script that I kick off whenever I add a data point, which in turn kicks off the other scripts like the example aboveYou'll notice that the example datafile has a "calibration datapoint" at the beginning. This causes the X axis of the graph to align correctly (read the perl script to find out why). Unfortunately, it also generates a false datapoint on the graph, so when you're looking at the graphs, ignore the first atapoint you see. The real data starts on the second datapoint.
