There certainly seems to be a lot of diversity in the levels at which the Scientific American books are written. While, for example, Sexual Selection by J.L. and C.G. Gould is very thorough and contains plenty of challenging ideas, this entry in the series is at the other end of the scale of complexity. Can you write for complete outsiders to a subject matter without talking down to them? No doubt, but Lewontin didn't always succeed in doing so.
In fact, the first chapter is very inauspicious indeed, very nearly condescending, let's-all-get-along-children along with social analysis on the level of Marxism-for-little-tots (especially noticeable since I just got through reading some much more sophisticated social and cultural history). Things do get better when Lewontin gets down to the actual science. The sections on basic genetics are very clearly written, though perhaps too long. The best thing in the entire book is chapter six, explaining some statistical concepts, and such ideas as variance and heritability. I have photocopied and saved those very clear explanations.
But there is a very serious problem with the book -- there are not only no "further reading" sections, but no references. Lewontin will state that "a study" said such-and-such, but not give an exact reference to the study. This would pass in a public lecture, but is not appropriate in a book. Again and again, Lewontin mentions only studies that support his statements, gives very little space to disagreements, does not mention current puzzles and unknowns, and makes it sound as if the subject is much more closed than it can possibly be!
I'm really frustrated with this -- such a contentious, active field of study deserves much more depth. Of course, the book is 25 years old now, so pretty out of date in any case.