If you have not written it up, you have not really done science in any useful fashion. If it isn't printed or posted, then it will need to be done again. Moreover, it must be reproducible, and how can some one test this ultimate criterion without your recipe?
Unlike many other genres (even scientific books), it is difficult to develop a great deal of comfort with writing original, scientific papers. The reason is simple. If you are being innovative, you are continually changing the subject about which you write. Each paper must be original, but the degree of originality required for publication varies greatly among journals. Be aware that negative results (failure to discredit a reasonable null hypothesis) generally do not merit publication without an explicit statistical power test demonstrating that failure is not simply a consequence of a weak test.
Scientific writing is set apart from other kinds of writing. You are trying to "open your skull" to let the reader know what you did and why. You need to be precise. You are generally writing in English but you need to be aware that many of readers will not be native speakers of that language. For all these reasons avoid mystery, pomposity, verbosity and metaphor.
Spend a great deal of effort revising the material up front (title and abstract), as they will be read many times and will determine whether the reader has any inclination to proceed deeper. Choose a short and informative title. Watch the syntax. Avoid jargon and abbreviations. Don't number a series of papers; it is presumptuous that they will all be published in sequence and that anyone would care. Use your first name, middle initial(s) and last name in the same form in all your papers; otherwise your scientific identity will be ambiguous. Write an "informative" abstract that briefly explains purpose, methods, results and conclusions. Avoid "indicative" abstracts ("... are described ..."). If in doubt (about whether something is important enough to be in the abstract), leave it out.
In terms of organization, don't try to come up with a new scheme better than IMRD (Introduction, Methods, Results, Discussion). This highly evolved structure is a boon to everyone, including the writer. The sole exception that you should consider is the combination of results and discussion when several diverse experiments or sets of observations are included and the combination allows a clearer and shorter presentation.
In the introduction identify and circumscribe the problem that you are addressing. Tell the reader why they should be interested as you review the literature, citing only the cogent reviews and highly selected, generally more recent papers. Nothing turns off a reviewer or reader more than an inappropriate citation; if in doubt, leave it out. Identify the general kinds of methods used. State the principal result and the principal conclusion from it. If the reader or reviewer is not engaged by this point, hope is lost. For that reason revise early and often. Each time you open the document, revise the title, abstract and introduction.
Give the methods and materials in enough detail that the results could be reproduced. Provide as much of the detail as feasible by citing prior work rather than repeating published description. In many professional societies, this section is considered critical to the extent that abstracts and other claims that lack the ability to reproduce them are not citable in journals produced by those societies.
What were the findings? If this results section is not short, you are mixing in methods and (or) discussion. Use a number of significant figures that matches the precision of your findings.
The discussion explains what the results mean and why they are important. Many first-time writers try to end this section with conclusions, which repeat information in the abstract. Unless the journal requires a separate conclusions end the paper on a single strong conclusion from your finding rather than a redundant listing of conclusions. Never pull a surprise here, that is, don't talk about issues and ideas in a conclusion that have not appeared earlier.
In terms of style,active voice often is clearest; work toward using it routinely but not exclusively. Deviate when doing so makes the writing clearer. Avoid "we" or "our" unless it is clear that you mean you and your co-authors. Speaking for an entire field or for non-authors invites disagreement. Avoid boxcars (strings of nouns parading as adjectives); once the number is greater than two, ambiguity and humor reign. Do not use "this" and "that" and other pronouns in place of nouns; as a kindness to non-native speakers put their antecedents in the same sentence. Do not repeat captions as text. Instead cite a table or figure parenthetically in a sentence that describes your principal conclusion from its data. Good practice in general (and in compliance with federal regulations regarding disabilities) is to describe in captions what you want the reader to see in the graphic or conclude from the table. The reader should be able to understand the graphic from the caption and figure legends alone. Use simple sentences.
Choose a journal by choosing your target audience first, then writing for them. Do so by writing for one person in that audience whom you know. Choose the journal by impact factor and use a society, not-for-profit journal when you can. Consider page charges, especially for long papers. Incessantly revise the title, abstract and introduction. Next most frequently revise the last paragraph of discussion and end on a strong note. You are known by your entrances and your exits. Consider making one figure in the paper general and interesting enough to be used in textbooks. Get informal review before you submit.
If the paper has weak points, highlight rather than hide them. Doing so leaves critics little to say and inspires confidence in your perception and humility. It also points to needed advances in the next research on this subject.
Day, R.A. 1994. How to Write and Publish a Scientific Paper. Oryx Press, Phoenix, AZ. 223 pp.
Strunk, W., Jr., and E.B. White. 1979. The Elements of Style. MacMillan, NY. 85 pp.