A crude graph of SARS over the past couple of weeks. Eaxt * = 50 cases, each x = 5 deaths

Date
Mar 17 ***
x
Mar 18 ****
xx
Mar 19 *****
xx
Mar 20 ******
xx
Mar 21 *******
xx
Mar 22 ********
xx
Mar 23 (no data)

Mar 24 *********
xxx
Mar 25 **********
xxx
Mar 26 **************************
xxxxxxxxxx
Mar 27 ****************************
xxxxxxxxxxx
Mar 28 ******************************
xxxxxxxxxxx
Mar 29 *******************************
xxxxxxxxxxx

I'm not an epidemiologist, but I would think the jump on Mar. 26 was likely a result of increased reporting, rather then an actual increase of cases. At this stage it's very difficult to get complete reporting as the disease is still new and it's symptoms a very similar, at least at first, to a numer of other diseases.

Note the dangers in extrapolating data, especially limited data. Taking the overall graph of cases between Mar 17 ad 29, there was an increase of almost 1400 cases, or over 800%. Extrapolating this 12 day period to 120 days would give something ridiculous like 13 trillion cases. Even worse, extrapolating the jump on Mar 25/26 all by itself would give 6 billion infections in 2 weeks, or 6 billion deaths in 17 days. Clearly an incorrect result.

In the meantime, CNN reports that SARS may be transmitted through the air, which would increase it's spread reatly (previously it was thought that it was only spread by direct contact or actual droplets from sneezes or coughs).