In a 1974 speech Richard Feynman famously described an example of confirmation bias in science:

One example: Millikan measured the charge on an electron by an experiment with falling oil drops, and got an answer which we now know not to be quite right. It's a little bit off because he had the incorrect value for the viscosity of air. It's interesting to look at the history of measurements of the charge of an electron, after Millikan. If you plot them as a function of time, you find that one is a little bit bigger than Millikan's, and the next one's a little bit bigger than that, and the next one's a little bit bigger than that, until finally they settle down to a number which is higher. Why didn't they discover the new number was higher right away? It's a thing that scientists are ashamed of—this history—because it's apparent that people did things like this: When they got a number that was too high above Millikan's, they thought something must be wrong—and they would look for and find a reason why something might be wrong. When they got a number close to Millikan's value they didn't look so hard. And so they eliminated the numbers that were too far off, and did other things like that...

I saw many publications reference this part of the speech, but to my surprise I have not been able to find the plot mentioned in the bolded sentence. I don't genuinely doubt Feynman's claims, but it seems fitting given the subject to ask for evidence rather than accept it blindly as an argument from authority!

Was anyone able to replicate this plot? How many data points could they find? Is the data as monotonic as Feynman claims, or did he omit some outliers for clarity?

  • 8
    Feynman is not implying the plot exists published somewhere. May 23, 2019 at 17:23
  • 4
    There is already an SE question about this: hsm.stackexchange.com/questions/264/… May 23, 2019 at 17:54
  • 1
    Looking at the numbers in the HSM thread, if anyone had confirmation bias, it looks like it might have been Feynman in this case. It seems like the next measurements are actually significantly larger, with measurements after that only showing slight changes. This seems more like what you would expect if the scientists had no real bias, and is purely the result of better experimental setup and analysis (i.e. having the right viscosity of air, or using a method that avoids that error).
    – JMac
    May 23, 2019 at 19:52
  • 4
    Having done the Millikan experiment once (50-odd years ago), I can believe that the new figures tracked as Feynman suggests. Much of the experimental result depends on the mathematical analysis done on the raw values, and it is easy to screw that up, or decide to discard values that don't "agree" with accepted values. I witnessed this in my class, where most students did the analysis wrong but got "better" values. May 23, 2019 at 21:41
  • 10
    This question shouldn't be closed. It's a notable enough parable in the sciences, so it seems like a good question for SE.Skeptics. If anyone finds the linked content a sufficient basis for an answer, then they can construct an answer based on it here.
    – Nat
    May 24, 2019 at 3:32

1 Answer 1



A much more accurate account is in THE CHARGE OF THE ELECTRON (1935):

In 1929 Prof. R. T. Birge (reference 1) published an acutely critical and masterly survey of our knowledge of the fundamental physical constants. It was a very timely summary, and it undoubtedly-if we may borrow from the vocabulary of another trade-did much to promote a desirable "constants-consciousness ” in the general body of physicists. This especially applies to two of the atomic constants, the charge e and the specific charge e/m of the electron.

At the time of Birge’s report, there were two distinct, and apparently equally well authenticated, values of e/m. Direct deflection methods gave 1.769 x 10^7, while spectroscopic methods gave 1.761 x 10^7 abs. e.m.u./ gm. As the maximum error admitted in each case was only of order 1 part in 1000, the relatively large difference between the two values was more than a little disturbing. On the other hand, Millikan’s value of e (only slightly modified in revisions of the calculations, made by Birge(1) and by Millikan(2) himself) was generally accepted as accurate to within 1 part in 1000. In fact it seemed very likely that the electronic charge lay somewhere between 4.768 and 4.772 x 10 ^ - 10 e.s.u. It is significant of the authority attaching to Millikan’s work that it was considered necessary to correct his values for the small difference between absolute and international electrical units and for a small change in the accepted value of the velocity of light.

The situation has oddly changed since 1929, the spectroscopic and deflection values of e/m now being in excellent agreement at something very near to 1758 x 10^7. There are, however, now in the field two values of e which differ by more than 7 parts in 1000; they bear in fact to one another almost exactly the celebrated and possibly significant ratio 136/137(3). The first of these is Millikan’s, the second is the value deduced from absolute X-ray wave-lengths by a method which was only beginning to be fully exploited at the time of Birge’s first paper.

where reference 1 is Probable Values of the General Physical Constants (1929) reference 2 is Millikan's THE MOST PROBABLE 1930 VALUES OF THE ELECTRON AND RELATED CONSTANTS (1930)

In other words, by 1929 there was data by a different technique that contradicted Millikan's value.

For further information see:

Note on the Value of the Electric Charge (1929)

Absolute Wave-Lengths of the Copper and Chromium K-Series (1931) (finds the e is 4.806 × 10^−10 e.s.u., above the currently accepted/defined value of 4.8032 × 10^−10 e.s.u.)

Viscosity of Air and the Electronic Charge (1935).

The charge of the electron (1937) (a follow up to the 1935 article with the same title)

The Atomic Constants A Revaluation and an Analysis of the Discrepancy (1939)

and see:

enter image description here

where the values found by both Bearden and Cork are above the current value.

So in 1931 it was first realized that Millikan's 1930 value disagreed with X-ray technique values due to Millikan's use of an inaccurate viscosity of air value. And this was generally accepted by the 1935-1939 time frame.

In conclusion, there was no hesitation to report correct values in the primary (experimental) literature, at most there was an initial reluctance in the secondary (review/commentary) literature to accept that the values by the new X-ray technique were correct.

  • Is that "1.769 x 107" (and odd value) or is it a tupo for 1.769 x 10^7? If so, the two values differ by less than one part in 1000. Oct 17, 2019 at 22:22
  • @DanielRHicks thanks, 1.769 x 10^7. I corrected most of the optical character recognition errors, but probably there are more
    – DavePhD
    Oct 17, 2019 at 22:43

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .