Question about T-Test and Size of Samples

I have a statistics question, and not being a statistician (I'm an
engineer :), I am a bit puzzled by the following problem. Any help
that could be offered would be greatly appreciated.

Let's say that I have two sets of data (with equal sample sizes, "n")
which are acquired from two (properly randomized) scientific
experiments. I have plotted these data, and for a large enough "n",
they take on a roughly normal distribution. As part of my
methodology, I've always employed a Student T-Test to evaluate whether
the two sets were drawn from populations with different means.

I've never really given this much thought because I've always had a
fairly large sample size. But, the issue that I'm currently facing is
that I am now getting sets with much fewer samples (e.g., instead of
~30 points of data per set, "n" may now be closer to ~5 points of
data), and am concerned that I may not have enough sample points to
"safely" draw any conclusions about the samples' means. A few rough
plots show that, with ~5 points, my data kind of look normal ...
but ... this is hardly a reassuring to me, mathematically.

This begs a question:

What is the correct way for me to go about determining whether or not
I have enough data points (in my sample sets) in order for me to be
able to use a T-Test to draw a conclusion about my population means?

Or, phrased differently: I want to be able to assess whether two sets
of data are drawn from the sample population; if the size of the sets
is small, is there some other kind of test that I should be performing
first to answer the question, "Do I really have enough results before
I can even make comparisons"?


regards,
Kris
.