Digitizing an old graph from Fleischmann

  • Here is a graph from:

    Fleischmann, M. The Present Status of Research in Cold Fusion. in Second Annual Conference on Cold Fusion, "The Science of Cold Fusion". 1991. Como, Italy: Societa Italiana di Fisica, Bologna, Italy


    This is Fig. 1:

    It is an important paper, and the OCR version of the text has so few errors, I thought I would do a complete conversion. In other words I am converting it to Microsoft Word and from that I will make a nice new Acrobat file. I got to figure 1 I thought, "this deserves close analysis." It deserves to be reset from scratch.

    I looked around the Internet and found a program that converts graphs into digital data:


    That is something many scientists want to do, I expect.

    Anyway, I have the program and I am learning how to use it. I have generated a bunch of data from the cell temperature curve in the graph, and then I generated this new x-y scatter graph from that data:

    I have not quite figured out how to use the program. This graph is missing the data points that drop down to 39°C at the beginning of day 4 and then rapidly recover. But you can see it is generally correct. Here is how the program sees Day 4 data:

    I can't seem to make those little purple spots extend below and then come back up. I'll figure it out sooner or later.

    The program is simple. You mark the starting and ending points and numeric values of the x-axis (2 to 6) and then the y-axis (38.5 - 40.0), and then click on the curve and the program generates a table like this:

    2.1109 38.5688
    2.1155 38.6207
    2.1204 38.6749
    2.1313 38.7224
    2.1505 38.7555
    2.1765 38.7746
    2.2041 38.7916
    2.2297 38.8145
    2.2530 38.8439
    2.2750 38.8776
    2.2958 38.9143

    . . .

    The first number is the x-axis coordinate, and the second number is the y-axis value for that coordinate.

    The x-axis numbers are not evenly spaced so you have to make it an x-y scatter graph (which my voice input program thinks is an "ex-wife scatter graph").

    There are some problems with the original image of this graph. The distance from Day 3 to Day 4 is longer than from Day 4 to 5 and 5 to 6. There is no marker for Day 2. So I estimated where it should be.

    I have not done the voltage yet.

    When I finish I will upload the new version of the paper and append this data table to it.

    This is somewhat off topic but:

    . . . The most famous example of digitizing a graph in the history of cold fusion was done by MIT. They manually added a bunch of data points to hide the apparent excess heat. See my illustration of this in a paper by Mel Miles, on p. 23:


  • The GetData program that lives in your computer might be a better choice. It may produce higher resolution. I have not compared it carefully, but today I did the following, which may be beyond the capabilities of the online program:

    I produced a 1200 dpi scan of the graph from the printed book.

    I used the GetData program to scan this image point by point, collecting ~4330 points automatically. That is about the limit to the program. I bumped it up to higher resolution but it only scanned half the graph.

    I manually added a few points that were missed at the ends of some of the spikes. Like so:

    Here is the result. This is the Potential section from the bottom of Fleischmann's graph:

    You have to sort the points into ascending order within the spreadsheet, then graph them as an x-y scatter graph. If you do not sort them, it comes out looking like this, because the points added manually go to the end of the list:

  • There is also a desktop version of WebPlotDigitizer, but it appears to be a web application and to be functionally the same as the online version:


    Robert Horst

    I found out about it a few years ago. It's still actively developed and it's been used and cited on actual research papers, which I guess gives the program author a moral incentive to continue working on it. I think it's also been cited in some LENR papers (for example here).



    It's possible to sort the data points within the program before exporting.

  • It's possible to sort the data points within the program before exporting.

    With the GetData program, that does not work well. It's a long story. Anyway, it takes only a millisecond to sort within the spreadsheet.

    Here is how my final version looks:

    Here is a close-up of the Potential curve new version (blue) superimposed on the original (black), for Day 1 and Day 2. This shows the limits of the point capture method:

  • As for accuracy, it does help to have a high resolution source image when available, and that it is free of paper or perspective distortion. Manual sampling is an option when automatic recognition fails; that takes much less effort with a pen tablet than with a mouse.