The website of the canadian newspaper "The Globe and Mail" seems to have a badly configured access policy of a log file. The log file is a standard web server log file that contains browser information, requested data and the IP address of each visitor that performs a search from the online edition of the newspaper. A simple test of this problem is searching for some know text (for example: "Hello World") using http://www.theglobeandmail.com (Globe 7-day Search) and few seconds later you will find an entry in http://archives.theglobeandmail.com/generated/Fragments/access containing the string "Hello+World" esteban.
