c0nc0rdance: No-one has ever been able to replicate Gregor Mendel's observations of pea plants. They're a little "too perfect", lacking even random statistical noise that would have been expected from small sample sizes. The big question: was it scientific fraud? (Art: Harald Ritsch)

No-one has ever been able to replicate Gregor Mendel's observations of pea plants. They're a little "too perfect", lacking even random statistical noise that would have been expected from small sample sizes. The big question: was it scientific fraud? (Art: Harald Ritsch)

Famed population geneticist RA Fisher published this paper in 1936 taking Mendel to task for either concealing, cherry-picking, or omitting parts of his study of pea genetics. There are three points of contention: digital.library.adelaide.edu.au/dspace/bitst...

digital.library.adelaide.edu.au

1. The segregation ratios (as in 'Mendelian ratios') are too perfect. Actual observations are modified by noise and distortion, only land on the 3:1, 1:2:1 ratios in extremely large samples sizes of ideal, perfect genetic models.

2. Mendel didn't observe linkage between traits that are actually linked. Sometimes genes physically located close together will always travel together, rather than segregating independently. Mendel somehow found genes that were 40% linked had NO linkage.

3. Mendel never did the F2 cross that would today be standard practice, and would have demonstrated even more clearly the deviations from random segregation. Fisher proposes this experiment was DONE, just not DOCUMENTED because it failed to follow Mendel's predictions.

More recently, Norman Weedon, a modern pea geneticist from Montana State, postulated some other ways that Mendel may have "streamlined" his documented results: methodological variables he didn't document.

For example, peas were collected in bundles in the Fall from each experimental garden, and assistants selected representative plants for Mendel's inspection. It's possible that an assistant was skewing the sampling to confirm Mendel's conclusion. (Art: Mikhail Dmitrievich Ezuchevsky)

I don't think the process of scientific hypothesis testing was mature enough at that point to call it fraud, nor was his research program formal enough. These days, it would be unacceptable. It is a very good teaching example

This. If he did it today, with the standards of the scientific method more clearly established it would be fraud. But if his intent was to correct "errors" in his experiments, I don't the method was established enough for him to know this was fraudulent. If he did it today, he should lose tenure.

Does this make Mendel one of the most important fraudsters in history?

Millikan did a similar thing with his famous oil drop experiment that established the charge of an electron. In both cases they had the data to prove they were right, but edited it to appear very, very right. Is it really fraud if it results in an accurate and predictive theory? It’s complicated

Well yes it is... it might not disprove the results but it means the report can't be trusted

That's it, I'm an anti-vaxer now

I'm anti-peas now. Loved them for their sentimental callback to Mendel's research but I feel so violated now. They are dead to me.

I too am now anti-P. Pity my poor bladder.

Ass the eas, lease.

Easy-easy!

www.instagram.com/reel/C3LHtkp...

Does this make you an esca-pea?

I've ghosted them so I guess I'm a Pea-Not.

i will not be appeased

It’s interesting to me because even learning it I was curious why there was not more variation. I figured it was a simplified version in order to explain the rough concept of genetics and how it would work. I’ll never look at peas the same again

This is a bit like calling out the Ancient Greek historians for including so many summaries of dreams instead of first hand accounts

But Mendel's own description is very thorough, involving e.g. a control group (because of insect damage), a discussion of which traits are suitable, and a detailed systematic account of the experiments themselves. If he did something different than he wrote, he's violating his own standards.

Reference: vlp.mpiwg-berlin.mpg.de/library/data...

VL Library Item: lit26745, p0001vlp.mpiwg-berlin.mpg.de The Virtual Laboratory - Essays and Resources on the Experimentalization of Life between 1800 and 1930

Seeing as at the time they didn't have quite our modern notions about how bad airbrushing the dataset is, I have no problem believing he or his team airbrushed. Remarkable that he airbrushed in a direction that supported useful theoretical inferences about non-observables, though.

It doesn't sound that much different from modern pollsters weighting polls.

@jonrosenberg.bsky.social has taught us about Mendel’s late life super villainy, so his less obvious crimes are no surprise to me.

???

Jon Rosenberg’s webcomic “Goats” had Gregor Mendel as a minor character. A minor character who was a super villain.

holy shit I haven't thought about Goats in years

@jonrosenberg.bsky.social

I used to have the first year PhD students in the IGC PhD program run the numbers and then had a "trial" with Mendel's defense team vs prosecution.

That's really interesting! Did you have them calculate the linkage values he should have gotten? What was the verdict, generally?

Varied year by year - the key starting text for the defense was Hartl & Fairbanks. Have a look, if you dont know it 👇 academic.oup.com/genetics/art...

Mud Sticks: On the Alleged Falsification of Mendel's Dataacademic.oup.com Anecdotal, Historical and Critical Commentaries on Genetics

It's fascinating to see the chain of "genetics legends" here that spans 4 generations: Fisher criticizing Mendel, Sewell Wright criticizing Fisher. Muddy waters to me. I'm glad your students get more into the weeds and analyze the primary source.

And nobody discussed his p-value ? 😉

I read this book ages ago when I first got into teaching science: www.amazon.com/Betrayers-Tr...

!!!

No, it wasn't; the Science was NEW, so there wasn't a 'body of evidence' to distort by adding 'bad' data. The concept of 'signal versus noise' wasn't well developed either; Mendel's task was to IDENTIFY a pattern, and he did. Finding it in a 'fuzzy' form would be a FAILED identification.

My own work illustrates the concept; I'm revising what's already (incorrectly) thought about Aerodynamics, so I can't render a 'precise' version of a Pneumofoil. Identifying it from 'first principles' would more likely produce a Platonic Ideal of the concept - either 'wrong', or PRECISELY right...

There is a very innocent explanation, and this accusation should be put to rest by now: doi.org/10.1093/gene.... TL;DR: when Mendel grew 10 plants it wasn't based on 10 seeds because he was a good gardener. He didn't mention this, and Fisher didn't think of it. Some traits show slightly in seedlings

Mud Sticks: On the Alleged Falsification of Mendel's Datadoi.org Anecdotal, Historical and Critical Commentaries on Genetics

it was God's hand ...

The story I heard (how could that not be conclusive evidence?) was that the gardeners felt sorry for the poor nutcase and made sure that he got the plants he wanted. So if it was fraud, he was the victim and not the perpetrator.

🤯

For another perspective, here's the last paragraph in Mendel's Wikipedia page:

It's only scientific fraud if he understood science the way we do. If he thought he was correcting "errors" so people would understand his observations & didn't think of this as wrong, he was operating under a different understanding of the role of a scientist. No fraud Definitely altered data tho'

The book "Betrayers of the Truth" also talked about this.

So, uh...what does this mean for genetics as taught in middle school? Is it all wrong?

No. It's right, but incomplete. Uncooked data would have proven it pretty well, but reality is a bit more complicated than his simple rules. His falsified data probably helped the study of genetics get started by making his "evidence" stronger.

Post