If you would like to comment on this week's Outlook articles, please write to the author c/o Outlook, The Washington Post, 1150 15th St. NW, Washington, D.C. 20071, or send e-mail to outlook@washpost.com. We would be glad to forward your response.

Comments for individual authors who have e-mail accounts can be sent to:

KENNETH COOPER (cooperk@washpost.com)

DAVID GOODMAN (goodmint@together.net)

ALLAN LENGEL (lengela@washpost.com)

CHARLES TRUEHEART (chartrue@compuserve.com)

GENE WEINGARTEN (weingarten@washpost.com)

If you are interested in having your comments considered for publication, please send them to our colleagues on the editorial page. They select and edit letters to the editor. They can be reached at Letters to the Editor (at the Post address) or by e-mail through The Post's Web site (http://www.washingtonpost.com). The Post's ombudsman, E.R. Shipp, addresses readers' concerns for all sections of the paper. She can be reached by voice mail at (202) 334-7582, and by e-mail at ombudsman@washpost.com.

CORRECTION: Several sharp-eyed Outlook readers pointed out an error in the hypothetical case we offered to illustrate a story about polygraph testing.

The mistake arose from the conditions we set. We specified an organization of 3,000 employees in which .5 percent of the employees were spies. We also specified the polygraph test had a "sensitivity" of .98 (which means 98 percent of the organization's spies will test positive).

Under these conditions, polygraph testing of the entire organization will identify 14.7 spies, and miss .3 of a spy. This outcome, while mathematically correct, is nonsensical in that it assumes there can be fractional people. Consequently, we rounded 14.7 up to 15, and .3 down to 0. (But did not note it in the graphic.) So it appeared that all spies in the organization had tested positive on the polygraph and that no spies were missed. That condition exists only if the test's sensitivity is 1.00, not .98.

If we had used a sample size of 10,000 employees, the number of overlooked spies would have been one, and the equation we offered to calculate sensitivity would have produced the specified .98 figure.