The abstract is prob­a­bly the most impor­tant part of a paper. Many read­ers will not read any­thing else, so you need to grab their atten­tion and get your main mes­sage across as clearly and suc­cinctly as pos­si­ble. It is not meant to be an intro­duc­tion to the paper, but a sum­mary of the paper. In a sin­gle para­graph, a reader can learn the pur­pose of the research, your gen­eral approach to the prob­lem, your main results, and the most impor­tant con­clu­sions. Write as if you have one minute to explain the paper to an inter­ested col­league, assum­ing that she will not read the paper herself.

A good abstract will con­tain the fol­low­ing elements.

1. What did you do?
2. Why did you do it? What ques­tion were you try­ing to answer?
3. How did you do it? State your methods.
4. What did you learn? State your major results.
5. Why does it mat­ter? Point out at least one sig­nif­i­cant implication.

An abstract should stand on its own, and not refer to any other part of the paper such as a fig­ure or table. Abstracts gen­er­ally do not have cita­tions either.

Other things to keep in mind:

• As a sum­mary of work done, it is always writ­ten in the past tense.
• Be explicit, pre­cise and concise.
• Stick to a sin­gle paragraph.
• Restrict back­ground infor­ma­tion to a sen­tence or two at most.
• Make sure that your abstract is con­sis­tent with what you reported in the paper. (This is par­tic­u­lar impor­tant when a paper has been revised and results may have changed.)

As a gen­eral rule, write the abstract last. After all, you can’t sum­ma­rize some­thing that is not yet written.