When we want to estimate the distribution of a random variable (e.g. particle diameter) in a number of particles, we may represent it either with number or area or volume(mass) frequencies. I know that hypothesis tests can be used to compare number-basis particle size distributions because the definition of a hypothesis test, for example Chi-square goodness-of-fit takes the frequency as a number count. I wonder if it is possible to use such tests when the frequencies are measured in mass rather than number of particles.