How to write a paper in physics? I really like to do research in physics and like to calculate to see what happen. However, I really find it hard to write a paper, to explain the results I obtained and to put them in order. One of the reasons is the lack of my vocabulary. 


*

*How do I write physics well? I think that writing physics is more dependent of an author's taste than writing mathematics is. 

*Are there any good reference I can consult when writing? 

*Or could you give me advice and tips on writing a paper? 

*What do you take into account when you start writing a paper? 

*What are your strategies on the process such as structuring the paper, writing a draft, polishing it, etc?

*In addition, it is helpful to give me examples of great writing with the reason why you think it is good. 

*Do you have specific recommendations?
 A: I will add, that nowadays the introductory part -and even the overall length of the paper- is more important that in classic times. This is because if you paper is too much specialist (and it will be) you must give the reviewers a hint that you have done your homework, that you known your field of study and that you can even give some pointers to guide the revision just in case that the reviewer is not working in your subfield. 
A: I bought The Art of Scientific Writing by Ebel, Bliefert and Russey a few years ago, and it's pretty good. However there is a huge amount that you can't really learn from a book. 
The first thing you need to do is to read a lot of papers. I can't stress how important this is. You need to know what is going on in the field and what problems are still open and which are closed. Even with the open problems you need to know what other people have been doing to try to tackle them. Ideally when writing your first few papers you would have an advisor or supervisor who is experienced in these things, and will help you in choosing problems, and with deciding how best to present the results. If you don't have this, then the importance of reading papers will be amplified again. So at first, read read read!
As you read papers you will start to get a good feeling for which papers are well written and which are not. It's fairly obvious, so you shouldn't really need us to tell you what we consider good writing. The style may vary from field to field as well, so giving you an unfiltered list is probably not very helpful. You really need to build up your own idea of what style seems most clear to you.
As regards actually writing a paper, the way I tend to approach it is to first write out the structure in terms of section titles (even for PRL type papers which don't actually use them, in which case I remove them later), then I try to break it down to the level of what I want to say in each paragraph or so. And then I start writing the actual content. This is just my personal approach, and is not going to suit everybody well. Then you proof read the paper, again and again to make sure that everything makes sense and that you have defined all the notation and ideas you are using before you use them, and you make sure the language flows ok, and that you haven't accidentally broken a proof (which is easy to do).
Personally I like a more didactic style, but that's not to everyone's tastes.
For your first few papers (and frankly any paper you consider very important) it is important to ask a few other people to proof read them. If you spend a lot of time on a paper, you become to close to the manuscript and often don't see errors or where it can be improved. If you are just starting out writing papers, this should be a more experienced colleague (someone who has written very many papers), and you should take their advice and/or criticism seriously. When starting out, at least, it is very easy to have a distorted view of your papers. You may also need advice on what type and level of journal you should submit to, and whether the preprint is yet of sufficient standard to upload to the arxiv (you also shouldn't be submitting to journals if the paper isn't good enough for the arxiv), and this isn't something you can learn via generic question on the internet. You need someone with experience in the area to read through the paper in detail, and give you unfiltered feedback on it. As you publish more papers, you are better able to judge these things for yourself, but at the start it is very easy to go wrong here.
Lastly, unless you end up writing everything on your own, which is extremely unlikely in physics (and not a good sign in my view) you will find that the style of the papers you write will often end up being some sort of mix or compromise of the styles of the various authors.
A: In addition to the Joe's answer, a bunch of good advices is here:


*

*G. M. Whitesides, Whitesides' Group: Writing a Paper (Adv Mat 2004), doi:10.1002/adma.200400767
Its two main points are:


*

*Start writing a draft as soon as you have some results, not - when the research is complete (as the later may never come). 

*Write in a way which is the most convenient to the reader, not - the writer.

A: I never forgot my old lecturer Robert Barrass and his book Scientists Must Write. - He never stood a chance, with me.
I still use the basic, 'Theory, diagram, experiment, results and conclusions' approach, otherwise I am lost!
A: References:
[1].  Joe Fitzsimon's response to this thread.
[2].  Piotr Migdal's response to this thread.
[3].  arivero's response to this thread.
Brief
Writing papers is no different to writing anything else, although with scientific papers, that has less to do with vocabulary than writing a novel, so you are not in such a bad position. However, that still leaves style, content and substance to be addressed, let alone how to keep your reader/reviewer reading.
The key, as with all writing, is to keep your reader interested.  The only difference with scientific papers is the context in which it is read, and the reasons.
Reformulating other Responses to this Question:
To reformulate other responses to this question [see the references], the phrase "keeping readers interested" means:
a.  Relevance:  Writing a paper that is relevant to other researcher's work.  As JFitz says, make sure your topic is current and not on a subject that has been closed. 
[Ref:  1]
b.  Standards:  Figuring out what is the standard for scientific papers in your field.  Hence JFitz's suggestion to read a lot of papers in your field.  If your paper doesn't match current standards, it will look unprofessional. 
[Ref:  1]
c.  Think Like a reader:  communication is all about being able to put yourself in the shoes of your reader.  You presumably know more about what you are writing than he does, so your reader is at a disadvantage.  You need to make the structure of your paper march in step with the development of the ideas. 
[Ref:  2]
d.  Language:  The language of physics is mathematics, so you can rely on this to convey your results.  However, the odd good analogy helps.  Just be careful of metaphors, they are pointless and irritating unless you are addressing laymen. 
e.  Peer reviews:  For learning to write novels, there are peer review sites, such as Authonomy.com.  I have never seen one for polishing papers, but there's an idea. 
[Ref:  1]
f.  References and summaries:  put people in the picture.  If you can't summarise what you are trying to achieve in a couple of short paragraphs there is something wrong.  The references give your reviewer/reader a handle and places to look for background information if they don't get what you are on about.  No paper is an island. 
[Ref:  3]
g.  Brevity:  Long paragraphs are boring unless you are Charles Dickens.  But then you wouldn't be writing papers... 
Conclusions:
The responses here say as much as they can to you.  Most people would work as part of the scientific community, and therefore, they don't need to ask these questions:  the institution they work for hammers it into them.  
But even if, nay, especially if you are working for some such institution, I hail you for making the effort to improve your papers.
If, on the other hand, you are working in the patents office in Bern, we will all be grateful for any extra clarity in your writing, and I hope that we have helped.
Help yourself by helping others is what this site is all about.
Parting Shot:
The reason for writing this response (apart from the two original contributions) is to illustrate that structuring what you write adds clarity and makes it easy to look up external references, as well as to help the paper be used itself as a reference.  
A: I find this one useful.
http://theory.tifr.res.in/~sgupta/edu/write.pdf
(Haven't read it in toto)
