Private User It is not common, but there are ways to start an isolated tree and expand that till you find the merging profiles to make the line complete again. Maybe it's wise to ask a curator to make a starting point with one of the older profiles and give you management, so you can continue your journey to their history.
The best thing to do is simply to document that unknown linkage in the About of your grandmother, particularly if you don't know the precise number and relationships of the persons "in-between".
Put in as much detail as you have (stories, etc.) which provides the anecdotal / oral history about that relationship, and then add a hyperlink to that 'known' 1720s profile -- all in the grandmother's "About".
On the other hand; if you *do* know all the relationships along the path (e.g.: X's father, son of Y, who married daughter of Z, who was son of <known-profile>), but no other details, then it is o.k. to add the explicit 'unknown' profiles along that path ; just don't "make up" information along the way.
Without a *real* connection, you can't. That is, however, an entirely different issue, which is addressed in another Geni Project about "Working with Relationships" (https://www.geni.com/projects/Working-with-Relationships/17570 ... and the discussions attached to that project).
In short, Geni only shows 1 of the myriad of possible relationships and the many possible pathways between 2 profiles. Furthermore, you currently don't have much control (some, but not much) over which path it shows you.
I REPEAT: Do not 'make up' information to support a presumed relationship.
Even if DNA patterns seem to support a relationship, DNA results cannot *prove* a specific set of relationships -- they can only (pretty thoroughly) dis-prove relationships. (Yes, DNA results can support the possibilities of some relationships over others, but the DNA data are only 'strongly suggestive' for close relationships.)