There are many books on scientific methodology, but guidelines for writing a good scientific paper are harder to come by. My month-long battle with a manuscript has made me ponder many aspects of writing.

First, there is the choice of scope. The best articles put their subject in a wider framework, but it's a fine line between relevant and ridiculous parallels. As a sociobiologist, I'm wary of making careless comparisons of ants to humans, despite frequent requests from friends both within and outside science.

Then there is the eternal temptation to promise more than your results deliver. My paper is on kin selection and insects. Do I suggest that it will change the way we study ants, the way we see social insects, or the roots of social interaction? The insanely inflated claims of some papers make me side with a friend of mine, who once suggested that articles should include only methods and results, leaving interpretation to the reader.

And how much room is there for speculation? Writing my paper has inspired many ideas about my ant system that remain untested. Gratuitous musings may transmit the ideas, but they make many reviewers see red.

After weeks of rewrites, I finally submitted my manuscript. At the moment, I have no clear picture of what it says or, indeed, why any of it matters. But I'm hopeful that given some time, I will be pleasantly surprised by my handiwork.