[Goose]

But the image manipulation here is that all these negative/"white" spaces were copied and then pasted. They don't show distinct separate experimental results. Which image is the copy and which is the original? Are any of them showing a real actual experimental result? We can't know.

Exactly so. It is concerning because there may have been "signal" in the lower right area that was pasted over with basically white space from the other green highlighted areas. Which way did the pasting go? We can't know. It is also interesting to speculate why there was any other "cut and paste" going on in this blot result. The other areas that are manipulated are not significant to the main result, so why were they altered at all?

It also seems to me that the whole figure is under-exposed compared to any of the other figures. In all the other figures the "no data" areas are light gray. In this figure they are basically white with some very weak "junk" present. The signal in the upper right corner is plenty strong though. While this "proves" nothing, it is interesting.

I believe the data that was sent to Cell had these blot photos in much higher resolution (1344 pixels horizontally??) than is present in the printed copies or available on the website. It is very clear why the reader (or MTL looking at galley proofs) would have seen nothing alarming, because "nothing" was what was seen and expected. I would argue that even with a higher resolution photo it would not raise questions. The area that is supposed to have no signal in fact has no signal. It only becomes suspicious when one notes that the "noise" in several areas are exact duplicates of each other. Nobody is going to be looking for that unless they are told to do so.

I have been told that "back in the day" sometimes gels/blots were cut to exclude areas of "no interest" to the study. This made them easier to store. If that was done here and then somebody had to "reconstruct" a middle, they might do it by copying and replicating data from each edge, lower right and upper left. Obviously, doing so would be improper, but it might "explain" what happened. Otherwise, what we see here makes no sense. If there was signal in the lower right you needed to conceal, copy over that and be done with it. Why mess with the rest?

Particularly when it comes to blots, this sort of falsification is very easy and thus concerning; you have to trust that the band or blank space without a band actually represents the protein or lack of protein the author tells you it represents.

Yep. You can also suppress faint bands by intentionally underexposing the photo. No photoshop required. Another data point is that this article has been cited over 850 times since it was published. That is a pretty big number I think. It may be that this particular issue isn't a driver for the paper being cited, but nobody has, at least AFAIK, challenged the conclusions from figure 3.

[Goose]

But the image manipulation here is that all these negative/"white" spaces were copied and then pasted. They don't show distinct separate experimental results. Which image is the copy and which is the original? Are any of them showing a real actual experimental result? We can't know.

Exactly so. It is concerning because there may have been "signal" in the lower right area that was pasted over with basically white space from the other green highlighted areas. Which way did the pasting go? We can't know. It is also interesting to speculate why there was any other "cut and paste" going on in this blot result. The other areas that are manipulated are not significant to the main result, so why were they altered at all?

It also seems to me that the whole figure is under-exposed compared to any of the other figures. In all the other figures the "no data" areas are light gray. In this figure they are basically white with some very weak "junk" present. The signal in the upper right corner is plenty strong though. While this "proves" nothing, it is interesting.

I believe the data that was sent to Cell had these blot photos in much higher resolution (1344 pixels horizontally??) than is present in the printed copies or available on the website. It is very clear why the reader (or MTL looking at galley proofs) would have seen nothing alarming, because "nothing" was what was seen and expected. I would argue that even with a higher resolution photo it would not raise questions. The area that is supposed to have no signal in fact has no signal. It only becomes suspicious when one notes that the "noise" in several areas are exact duplicates of each other. Nobody is going to be looking for that unless they are told to do so.

I have been told that "back in the day" sometimes gels/blots were cut to exclude areas of "no interest" to the study. This made them easier to store. If that was done here and then somebody had to "reconstruct" a middle, they might do it by copying and replicating data from each edge, lower right and upper left. Obviously, doing so would be improper, but it might "explain" what happened. Otherwise, what we see here makes no sense. If there was signal in the lower right you needed to conceal, copy over that and be done with it. Why mess with the rest?

Particularly when it comes to blots, this sort of falsification is very easy and thus concerning; you have to trust that the band or blank space without a band actually represents the protein or lack of protein the author tells you it represents.

Yep. You can also suppress faint bands by intentionally underexposing the photo. No photoshop required. Another data point is that this article has been cited over 850 times since it was published. That is a pretty big number I think. It may be that this particular issue isn't a driver for the paper being cited, but nobody has, at least AFAIK, challenged the conclusions from figure 3.