Jet spaces for maps with constraints

Lets be in the category $\mathbf{M}$ of smooth finite dimensional manifolds with smooth maps:

Suppose we have the set of all smooth maps $Hom_\mathbf{M}(R^n,M)$ from $R^n$ to a smooth manifold $M$. Taking only the first Taylor-Approximation of any such map, gives us the n-velocity Space $T_nM$, that is a fiber bundle over $M$ and a generalization of $TM$.

Now suppose we just have a subset of $Hom(R^n,M)$ given by some constraints on the maps lets say $$Hom_\mathbf{M}( R^n,M | constraint_1(f) ... constraint_n(f) )$$.

Does this always gives us a subbundle of $T_nM$?

If not, what should be required to be a subbundle?

Or say it the other way around: What must we proof, to show that the appropriate Jet set is a subbundle of $T_nM$?

(For my purposes its enough to suppose, that the constraints are of zero-order i.e. don't involve any derivations)

-
Could you say a bit more about the nature of the constraints? Are they algebraic equations in coordinates? – S. Carnahan Aug 20 '11 at 10:13
Yes, they are given as algebraic equations involving the $(x_1,...,x_n) \in \mathbb{R}^n$ and the functions of course. – Mirco Aug 20 '11 at 10:30
... Things like f(x,x)=f(-x,x). (This is just an example not a definitive required constraint) – Mirco Aug 20 '11 at 10:33
I've never heard the term "n-velocity space" before. What is the precise definition? – Deane Yang Aug 20 '11 at 13:14
I'm pretty sure the $n$-velocity space is isomorphic to the $n$-fold fiber product of $TM$ over $M$, or equivalently, the tensor product bundle of $TM$ with the trivial vector bundle of rank $n$. – S. Carnahan Aug 20 '11 at 16:01