It would be convenient if cf.histogram had options to normalise the result by (a) dividing by the total of counts, giving probability in each bin, or (b) also dividing by the width of each bin, giving the probability density function. Please excuse me if I have overlooked an existing method for this.