Question 1141128
want a t-test but estimate first with a z, since the z (0.995) is the same regardless of the sample size, whereas the t-value changes.
error=t*s/sqrt(n) use 2.576 for z in place of t 
1=2.576*2.4/sqrt(n)
square both sides and multiply through by n
n=38.22 or 39
now try tdf=38,0.995 which is 2.72
2.72*2.4/sqrt(39)=1.04, too big.
and with a few iterations, a sample size of 42 will be found to give an error of 0.9998, under 1.