A lot of static pages aren't being cached because they are set to expire immidiatly in the http header.

For example, when I try to open this page
GET http://jira.atlassian.com/browse/JRA-9448 HTTP/1.0 => HTTP/1.1 200 OK [1.453 s]

these 6 files are all set to expire immediately which forces the browser to get them for every issue I request.

GET http://jira.atlassian.com/styles/combined-printable.css HTTP/1.0 => HTTP/1.1 304 Not Modified [0.312 s]
GET http://jira.atlassian.com/styles/combined.css HTTP/1.0 => HTTP/1.1 304 Not Modified [0.438 s]
GET http://jira.atlassian.com/includes/js/combined-javascript.js HTTP/1.0 => HTTP/1.1 304 Not Modified [0.266 s]
GET http://jira.atlassian.com/includes/js/calendar/calendar.js HTTP/1.0 => HTTP/1.1 304 Not Modified [0.687 s]
GET http://jira.atlassian.com/includes/js/calendar/lang/calendar-en.js HTTP/1.0 => HTTP/1.1 304 Not Modified [0.438 s]
GET http://jira.atlassian.com/includes/js/calendar/calendar-setup.js HTTP/1.0 => HTTP/1.1 304 Not Modified [0.391 s]

So basicly each request is taking 4 seconds and 60 % of the time is wasted on retrieving Not Modified responses

see also latest comments for JRA-9448

This makes jira almost useless on really slow network