publications - Abbreviations (abbrv.) in journal (jrn) articles (artcs)

As we all know, you typically abbreviate something the first time that particular term shows up in a paper. However, I've been told in the past that certain things in a paper should be written in such a way that 'it stands on its own'. One such thing would include the abstract of the paper. So, perhaps a appears in the abstract that I want to abbreviate. Now one of two things can happen.

1.) If appears more than once in the abstract, I could abbreviate the first instance and then just use the abbreviation from that point on.

2.) If appears only ONCE in the abstract, then I shouldn't abbreviate it because then I would have defined an abbreviation that I would not have used again (given the concept of 'the abstract should be able to stand on its own).

The problem with (2) is that people will immediately think that the 'first instance' of should be abbreviated and don't immediately consider the 'stand-alone' idea so I'm either having to revise or try to put up a convincing argument for what I did.

Also, I've been told by some that tables and figures should be able to 'stand on their own'. So, if the abbreviation for appears in the table somewhere (or the caption), then I should explicitly spell out and then define an abbreviation for it right then and there, even if I have already done so earlier on in the text of the paper.

So, my question is, what is the proper way of handling these abbreviations and is the idea of 'this piece must be able to stand on its own' valid (and if its valid, what exactly does this idea apply to)?

---

Additional Information:

I am a computational chemist so we pretty much are forced to use the alphabet-soup of acronyms. MP2 is preferred over "second-order Moller-Plesset perturbation theory" and is commonly used. CCSD(T) is preferred over "The coupled-cluster method that includes all single and double substitutions as well as a perturbative treatment of the connected triple excitations". The list can go on and on. If this wasn't abbreviated in an abstract, having to 'spell it out' even a few times would make for an incredibly long abstract, figure caption, table, etc.

Answer

The first comment will be, do not abbreviate just because something occurs more than once. Abbreviations other than established ones (within your field) such as DNA, EDTA make reading more difficult. Of course all established abbreviation were new at some point but the message is, be restrictive. I understand your field may be in need of many abbreviations so make adjustments to these general comments accordingly.

Now as for the abstract, I would recommend to not abbreviate anything even if it occurs more than once or twice (again barring established abbreviations). The abstract should be seen as a separable part which is (hopefully read) by a wider audience than the paper itself. If you need to abbreviate something in the paper, do so in the main paper as the "first occurrence".

Tables and figures should be made to stand alone if possible (which probably is 80+% of the time). Often figures and tables may be the parts others take up when they describe your work. To have self-explanatory figures and tables is thus useful. With a table the table caption should be an integral part so I think it is reasonable to have abbreviations in the table body as long as the abbreviations are explained in the table caption. The same could apply to figures as well but I would go further and aim for making the graphics along self-explanatory even without its caption.