A radioactive source emits two kinds of uncorrelated radiation A and B. This radiation is observed through a counter which is able to distinguish A from B. In a given time interval, the countings are NA=200 and NB=1000. What is the statistical uncertainty on the ratio R=NA/NB? In case the observation time were reduced by a factor of 250, do you think that the new ratio r=nA/nB would represent an unbiased estimator of the true ratio between the two activities? In case of negative answer, how would you correct for it?