Oooooh boy, someone (actually multiple someones) just committed PD in a BIG way. 'Make up' is a rather ambiguous phrase. It could mean 'fabricate out of nothing'...or it can mean 'construct', as in 'Before I left the house, I made up my bed.'
Someone published an article in a scientific journal. In that article, there's a simple line of text that somehow wasn't caught by the primary author, the primary author's boss, or the EDITOR of the journal. That simple line of text instructs one of the authors to "make up an elemental analysis"- as in, fake the results.
Thats an extremely charitable view of the situation. I'm not an expert on clinical trials or articles, but I have read a quite a few due to a past job. Typically if data that exists is supposed to be inserted in to an article. The actual location of the data is given. e.g. insert data from trial abc-123 or insert data from [specific article]. An actual scientist could explain it much better than I.
In the context, it pretty clearly means, "invent out of whole cloth." The elemental analysis doesn't match the compound they're supposedly describing. So, not only did they fabricate it, they fabricated it BADLY.
I don't know they journal, but I assume it's peer reviewed. How the heck did 3 reviewers, the EDITOR and the copy editors miss such an egregious error? I actually once had someone send me in what was clearly the wrong version of a manuscript (I was co-editing a book at the time). It had about 6 different colors of track changes in it, along with comments among the various authors. I rolled my eyes and sent it back to them to fix. They were suitably embarrassed.
I've published and reviewed a few papers and so I can answer how this could happen. The incriminating section "Emma, please insert NMR data here! where are they? and for this compound, just make up an elemental analysis…" was not in the main body of the paper, but was in the supporting info. It used to be that supporting info (which often is much longer than the original paper) was never published and was only available if you requested it from the author. Now it is always available with the paper, but only someone who really, really cares will read it. There is rarely anything interesting there; it's more of a glorified blibliography. You might check it if you make the same compound and want to check the nmr data etc...
Anyhow, the reviewers job is to make sure the science is good (generally this means interesting, novel, significant etc..). But the supporting info doesn't tell you anything interesting so as a reviewer you might not look at it or look at it in any detail.
However, the main thing is that it is a separate file than the paper itself.
Author: Paper and Supporting Info submitted to website
Editor: Looks over and rejects outright or sends to review
Reviewers (2-3): Look over and give comments, edits and recommendation (accept, accept with revisions, reject) note: reviewers are anonymous to author, but known to editor
Editor: Looks over comments and sends reviewers comments to author with editor's decision (eg: accept with revisions)
Author: Revises paper; may include new experiments.
Author: New Paper and New Supporting Info submitted to website
Editor: looks over (may send to reviewers again, but unlikely)
Editing staff: Proof the paper (comma, spacing, abbreviations etc..) May or may not proof the supporting info.
Paper and Supporting Info published.
So, even though multiple people seriously, seriously screwed up, I can see how someone could upload the wrong version of the supporting info late in the process and no-one would notice.
I was part of a group that completely failed to upload the supporting info in the initial run through and then wondered why all the comments referred to info that was clearly laid out in the supporting info. (It wasn't really anyone's fault. There were multiple files. They all got emailed to the person who was uploading. Then 3 of the 4 files got minor edits and were resent so the uploading person only uploaded the 3 new files )
Another time, a friend got a paper where all the comments on the side were left with the discussion between the authors.
None of those involve fakery of course but I can see how these mistake can happen.