If you know (or can make an educated guess) as to the variability in your results, then there are good applications on the internet for calculating sample size.

Are there ethical issues? If you must kill the animals to get the data then that will limit your sample size. If you have to artificially infect the lambs with the nematode and the nematode causes considerable suffering then there might also be ethical issues that will limit your sample size.

Given that there is no prior information to enable use of sample size calculators, and there are no ethical issues, then make the sample size as large as your budget will allow. Having too much data is seldom a problem. Having too little data is a huge problem. Also if you plan on a sample size of 500 per breed, and farmer Joe goes out of business taking half your experiment with him. You may still have enough data with only 250 individuals per breed. Also consider that your data here can help others make better choices. If you take a sample size of 5 lambs/breed you will have considerable error in your estimate of variance. It will not be possible for others to know if 10 lambs were sufficient or if you really needed 500. If you use 500, then it will be easy to justify use of 40 next time. Furthermore, the probability that you will end up concluding that you had an insufficient sample size will be greatly reduced.

I would also check the literature to see what minimum sample size is accepted. This is not quite the same as what you should use, but it provides a measure of the minimum level of work needed to perform acceptable research in your field of study.

Thank you very much Timothy,

The trial is under natural infection during grazing and according to flock size (200 katahdin and 200 Pelibuey) we don´t have problem with ewes. But in the case of lambs considering the reproduction management we only have 20 lambs per breed. So, we want to be sure that sample size is enough.

Try playing with this:

Num=20

SD=5

REPS