A finite repetition-free sequence in a set is an injective function from an initial segment of the natural numbers into , including the empty sequence for . For a finite set of size , the number of these sequences is
The order of each list is part of the data, so it can be enumerated without choosing an order on its underlying support.
If is a Dedekind-finite set, its set of finite repetition-free sequences is also Dedekind-finite. An injective sequence of distinct finite lists would, by countable union of explicitly ordered finite lists without choice, either produce an injection or use only finitely many entries. The latter possibility is impossible because a fixed finite pool supports only finitely many repetition-free lists. If is infinite, the one-entry lists also show that the resulting set is infinite.
Let denote the finite repetition-free sequences, including the empty sequence. The map is injective, so this set is infinite when is infinite. Suppose, for a contradiction, that it contains a countably infinite subset. Fix its given injective enumeration ; this is part of the supposition, not a choice from a family of sets.
Enumerate all pairs with by their natural-number pairing codes, recording the corresponding entries . If the union of entries were infinite, recursively selecting the first new entry would give an injection , contradicting Dedekind-finiteness. Therefore the union is a finite set , say of size . A repetition-free sequence over has length at most , and there are exactly
such sequences, with the empty product equal to one. All would belong to this finite set, contradicting their distinctness. Finite counting and the least-code construction use no axiom of choice. This proves finite repetition-free sequences preserve Dedekind-finiteness and the required infinitude:
Write for the recursively repetition-free labelled trees. Each rooted tree is finite because the inductive definition forms it from an already constructed finite list of finite child rooted trees. Its label list is obtained by visiting the root first, then each child subtree in its prescribed order, using a depth-first traversal of a tree. This list can have repetitions. In particular, different branches may use the same label; the children are required to be distinct rooted trees, not to have distinct root labels.
For any finite label pool , prove by mathematical induction on that is finite. For it is empty. For each root , the children form a finite repetition-free sequence from , which is finite by mathematical induction. There are therefore finitely many child lists for that root, and the finite union over is finite. More explicitly, if is the number of rooted trees on a pool of labels, then
There is no countable choice in this finite mathematical induction.
Now suppose were an injective enumeration of a countably infinite subset of . Their canonical traversal lists give an explicit enumeration of all labels used. If infinitely many different labels occur, the least-first-occurrence procedure gives a countably infinite subset of , a contradiction. Otherwise all labels belong to one finite set . Every then belongs to : recursively its root lies in and its child rooted trees use only with that root removed. But is finite by the preceding mathematical induction, again a contradiction. Finally the single-vertex rooted tree labelled is injective, so is infinite. Hence recursively repetition-free labelled trees preserve Dedekind-finiteness:
For a label set , define inductively: choose a root label and a finite repetition-free sequence of rooted trees in as its children. Each object is a finite ordered rooted tree. Labels are distinct along each root-to-leaf path, and child rooted trees at a vertex are distinct as whole rooted trees. Labels may repeat across different branches; children need not have distinct root labels. For a finite pool of size , the exact number of rooted trees satisfies
This follows by choosing the root and then an ordered repetition-free list from the finite pool of smaller rooted trees.