I have a version that goes by week, too. The data previously (not updated) was:

4/29: 148

5/7: 2,223

5/14: 4,126

5/21: 4,537

5/28: 4,476

6/3: 3,763

6/11: 9,501

Certainly a very long period with flat growth, punctuated by a double at the end. Will be interesting to see if it did double again for both 6/18 and 6/25.

]]>If the total cases grows exponentially, then yes, its rate of increase and all further derivatives will be exponential. But even if the disease spread were perfectly exponential, and all noise due to measurement/sampling difficulties (eg cases unreported, or confirmed cases delayed declared in clusters) then the daily new cases would be expected to be noisy due to small number of cases (per day).

To see how surprisingly good the fit to exponential growth has been over the last month or so, see the wiki page:

http://en.wikipedia.org/wiki/File:Influenza-2009-cases-logarithmic.png

As of now (July 26), all major graphs have been closely exponential since May 14th or so. (thus appearing as straight lines on a semi-log plot)

]]>