It is a rice genotypes evaluation and I want to use 1000 grain weight, but these genotypes can not produce 1000 grain yield due to drought so I want to scale down to a 100-grain weight.
Dear Charlie Suruban considering your situation; it would be worth divide the whole sample in several subsamples (100 seed could be a suitable quantity), and calculate the average for individual/genotype. In this way, you'll reduce the effects of intraplant variations, and the microenvironment errors.
You can also take 100-grain weight, but nowadays researchers are taking 1000-grain weight very frequently for contrasting rice genotypes with superior grain quality.