The following are just a few notes on manuscript or student report writing for my area of science, which I would circumscribe as plant systematics, biogeography and evolutionary biology. I will probably at some point use this or a revised version for other purposes, but thought it might be interesting to blog about.
As should be well known, the typical research paper (and a student report mirrored after it) in my area has, in this order, a title, author names and contact info, an abstract, key words, introduction, materials and methods, results, discussion, optionally conclusions (or they may be part of the discussion), acknowledgements, reference list, figure legends, tables, and potentially appendices or supplementary data. I will only deal with some of these for now. The main text, i.e. without reference list, should probably not be longer than c. 5,000 words, and the shorter the better.
The abstract should summarise all the other sections in really abbreviated form: what is the paper about, what main methods were used, what are the main results, and what do they mean. Do not cite references in the abstract, and do not provide taxonomic authorities after plant names. They are provided on the first mention (and only on the first mention) of a name in the main text.
This section provides background information, describes the question or problem, and ends with aims of the study. Ideally start with general, well-established, and unproblematic claims ("biodiversity loss is accelerating", "genomic data have become increasingly available for use in phylogenetic studies", etc.) and move relatively quickly to the problem ("guidance is needed to prioritise conservation planning", "but analysis of the large amounts of data produced by high-throughput sequencing is computationally challenging"). Do not begin with your study group, unless the paper is a purely taxonomic or phylogenetic one, as this will not draw in as many readers; the introduction of the study group should come after the general question has been established.
Obviously the introduction needs to be full of references, and all but the most widely accepted statements should be followed by at least one of them.
The aims should follow logically from the questions or problems and need to tie in logically with the methods, results and discussion. They can be formulated as hypotheses or questions, but should not be too vague. Think "we will test if patterns agree with those postulated by Smith (1980)" instead of "we want to explore the patterns".
Be as concise as you can. Cite software, tests and methods, but not the most basic ones; PCR or t-tests, for example, can be considered sufficiently established. Start with the materials (sampling strategy etc.), then work logically through the lab and analysis pipeline.
This section presents the results and nothing more. Any sentence that interprets the results or explains them does not belong here but into the discussion. References do not belong here. Explanation of how something was done does not belong here but into the methods.
On the other hand, all observations need to be stated explicitly in the main text of the results section, with a reference to the figure or table where they are presented. The figure legend itself should, in turn, be very concise and merely provide enough information to understand the figure, but it should not restate what the reader can see in the figure just above said legend anyway. For example, the figure legend might say "A, map of Australia showing endemism hotspots", but something like "hotspots are found in the southwest and southeast" is superfluous here and goes into the main text: "endemism hotspots are found in the southwest and southeast (Fig. 2A)".
Perhaps the hardest part to write, it explains what the results mean and how they fit into the wider context. It may also end on suggestions for further study. The main problem for beginners is to avoid repeating what was already said in the introduction and results sections.
The discussion should again have lots of references - not because we always need to cite lots of papers per se, but because any paragraph in the discussion that does not have at least one reference can be considered under suspicion of either belonging into the results section or being mere padding.
Collaborators, funding sources, peer reviewers. For student reports, one would expect the supervisor(s) to be mentioned. It is not clear to me why many people feel the need to initialise the names of those who they are thanking, as writing them out makes it easier to identify them, especially if they have common family names.
I haven't really made a tally, but the typical article would probably have between two and six figures, after that it becomes excessive.
Journals in my area expect the manuscript text and figures to be uploaded as separate files, and they expect high quality file formats such as EPS for vector graphics and TIFs with lossless compression for bitmaps.
For exchanging a manuscript draft between co-authors, or a student report draft between student and supervisor, it is, however, probably best to insert figures into the manuscript file, because then your collaborator has less files to handle and to print. I would suggest the following approach:
Have a separate section for the figure legends after the references. This is as it should be for journal submission.
Produce EPS (vector) or JPGs (bitmap) of the figures to save on file size and insert each of them into the text directly above the relevant figure legend. Although Word allows for them, I would very strongly advise against using convoluted text boxes-inside-object boxes-inside-object boxes. A student report I once had to reformat froze my computer for several minutes when I tried to resize one of its "figures". Ultimately I had to export the page as a PDF and then export the PDF from Inkscape into yet another format, whereupon I reinserted the figure into the Word document. A journal would, of course, not accept anything like that anyway.
Select the wrapping option "in line with text". This will treat the figure like a character, meaning that it will remain anchored to its position relative to the text no matter what you do upwards of its position. If you use other wrapping options such as "on top of the text" the figures will float around freely, and changing line spacing or font size, or deleting paragraphs, will mess everything up to no end. I once dealt with a student report where three figures ended up on top of each other!
Do not use text boxes for the figure legends, or for tables, for that matter. Really, why would you? They can be normal text, just like the rest of the manuscript. In fact I have yet to see a use case for Word's text boxes in my line of work (PowerPoint is a different matter).
There is no reason to use words like "whilst" where a simple "while" will do.
Often sentences can be simplified greatly; some people seem to have a penchant for writing something on the lines of "for pollination, it has been demonstrated that hummingbirds are more efficient", but the same could be said as "hummingbirds are more efficient pollinators" - saved us six words! Similarly, "in the literature it is documented that the sky is blue (Smith, 1980)" can be reduced to "the sky is blue (Smith, 1980)". And yes, I have seen sentences like these, particularly as a reviewer.
Do not use double spacers at the end of a sentence.
Do not have spacers at the end of a paragraph.
There needs to be a spacer between any measurement and its unit (5 km, 5 h), with the exception of temperatures (5ºC) and angles (5º).
This might seem a bit OCD, but believe me, you do not have to make some poor copy editor's life harder than it is.