What to do if the authors made a typo?

[danieleongari]

Nitrogen isotherm at 77K in HKUST-1 10.1016j.cattod.2012.03.034.isotherm1 results in a pore volume >10x the expected one. A direct investigation of the paper (https://www.sciencedirect.com/science/article/pii/S0920586112001940?via%3Dihub) revealed that the person in charge of the digitization did his job diligently. But this was not the case for the authors, who wrote mmol/g instead of cc(STP)/g: this second one would be consistent with their reported pore volume.

I wanted to rise this case, which is basically impossible to spot at digitization, unless running some subsequent analysis that could spot some weird order of magnitude in the amount of adsorbent per gram. I would suggest to correct anyway the data, when the correction is so obvious, and maybe mention the correction in a "comment" section of the json.

[danieleongari]

Thinking more about making a correction, the problem is that I can tell that it needs a fix but I can not tell if the right one is to change the units or if the authors forgot the dot in the values which should be in the 30.0 - 35.0 range. This could also be reasonable.

[dwsideriusNIST]

@danieleongari In a case where there is an identified fix to an isotherm, I take editorial privilege and correct the data. But I have to be certain of the correction, either via correspondence with the author(s), strong thermodynamic grounds, or based on experience. The case you identified falls in the 'strong thermodynamic grounds' category. (An example of experience: if a hydrogen isotherm at 77K is plotted versus p/p0, I know that p0=1atm. It's an unforgivable mistake by the authors, but 100% of the instances that I've investigated have confirmed p0=1atm - consequently, I longer contact authors about this.)

I now have an 'audits' table in the database, which will eventually be an output option in the API. One of the columns in that new table is for comments where corrections like this can be annotated.

Good catch, though, as this isotherm is a gray area where the digitizers did their job but the authors did not. I ran my own BET analysis on the isotherm (which BTW, this isotherm doesn't satisfy the Rouquerol criteria at all, so caveat emptor). My area calculation came to 970 m2/g by swapping cm3(STP)/g for mmol/g. I did your suggested transformation where the y-values were divided by 10 and then taken to actually be mmol/g, but then obtained 2176 m2/g, which is out of range for the BET area of a good CuBTC sample in my experience. Have I done some math incorrect here?

My instinct is that the units should be cm3(STP)/g, without a more complicated mistake.