It is desired to compare the compression resistance of three concrete mixes. The "normal" procedure is to make samples in duplicate. You consider determining a sample size that allows you to establish differences when they are above 25 kg / cm2, with a power of no less than 0.90 and a probability of committing type I of 0.05. It is known from previous tests that the standard deviation of the resistance is 7 kg / cm2. Is the procedure to duplicate correct? How many replicas would you recommend and why?