1: \begin{definition}

2: A \textit{subgerm} of a germ $\gamma$ is the result of forgetting an arbitrary number of its arrows, in such a way that every graph corresponding to a polygon with more than two leaves remains connected -- although two-leaved polygons may completely disappear. This condition means that subgerms must remember the codimension of $\gamma$ -- \textit{but the Gauss degree may drop}.

3: 

4: We set $\I(\gamma)$ to be the formal sum of all subgerms of $\gamma$ that are signed tree diagrams. It is understood that subgerms are counted with multiplicity if the removal of distinct sets of arrows yields homeomorphic results. This defines a linear map 

5: $$\I: \mathcal{G}_i \rightarrow \mathcal{T}_i.$$

6: If $\tau$ is a signed tree diagram, we define $\T(\tau)$ as the underlying tree diagram, multiplied by the \textit{product of the signs of all arrows of two-leaved polygons}. Again this extends into a linear map 

7: $$\T:\mathcal{T}_{i,d} \rightarrow \mathcal{A}_d^i.$$

8: 

9: \begin{remark}

10: The fact that the map $\T$ disregards the signs of arrows associated with polygons that have more than two leaves should be interpreted this way: for these polygons, the signs of the arrows have already contributed by entering the co-orientation defined by germs on their associated strata. In other words, when a simple crossing merges with others into a multiple crossing, we stop regarding its writhe as making sense individually. See Lemma~\ref{lem:epsi} and Theorem~\ref{thm:top}.

11: \end{remark}

12: \end{definition}

13: