I have read a children's book where alien race of "square people" used a pair of compasses that drafted a perfect square when used. Now I wanted to explain to the child that it is not possible to have such a pair of compasses, but then I was not really sure. So is it possible to construct a mechanical tool with one fixed point and one moving point that would actually draft a square with one continuous movement?
It depends on how restrictive you are about what counts as a 'pair of compasses'. For instance, if you require the writing end to be a fixed (Euclidean) distance away from the fixed point at all times, then that pair of compasses will always draw a circle, since a circle is exactly the set of points at some fixed distance from a fixed point.
One way you might be able to build such a pair of compasses is by using linkages. A linkage is like a more complicated version of a pair of compasses, where you have some collection of fixed points, movable joints, and bars connecting them, with a writing stylus attached at some point. A pair of compasses, then, can be viewed as an extremely simple type of linkage. A very famous example of a linkage, dating from 1784 is Watt's linkage, which was used in the first steam engines. It is special because it draws a nearly perfect straight line:
James Watt once said that of all his inventions, this linkage was the one he was most proud of. It wasn't until 1864 that a linkage was discovered that could draw an exact straight line. It is called the Peaucellier-Lipkin linkage, and earned its inventors the Prix Monyon:
How do you get from here to a square? The following remarkable theorem was proved in 1994:
Theorem (Thurston, Kapovic-Millson): given any polynomial map, there is a linkage and a vertex of that linkage such that the vertex traces out that polynomial map.
In this context, a polynomial map is any curve on the plane that can be described by polynomial equations. Certain curves (like $y=\log(x)$) do not have polynomial maps describing them; however, any curve can be approximated to arbitrarily high accuracy by polynomial maps. We therefore have two nice corollaries:
Corollary 1: There is a linkage that signs your name!
Corollary 2: There is a linkage that draws a square (to arbitrarily high accuracy).
Bear in mind that the theorem does not guarantee that the linkages will be simple! In fact, if you wanted to build a linkage that actually did sign your name, it would probably be extremely complicated! But it would nevertheless exist. I have no idea how complicated a linkage would have to be to draw a square.
Further reading: the reason I know about all this is that I've read Joseph O'Rourke's excellent book How To Fold It, which has a website here. If this kind of thing interests you, I strongly recommend buying a copy.
You might want to explore this link about drilling square holes. Your "central point" constraint goes - a curve of constant radius is a circle, but a curve of constant diameter need not be.
Does the book actually describe its use as "One continuous movement"?
If not, it seems likely the "pair of compasses" in your book is not referring to any exotic contraption. "Pair of compasses" = "compass"
When used correctly, (with a pencil and a straight edge) you can easily draw a perfect square.
(draw lightly until step 12, most marks will be temporary)