This is a distinctly difficult passage to make out clearly or with any degree of certitude.
GP Baker & PMS Hacker, Wittgenstein : Understanding and Meaning, are less than lucid :
'Zwischenglieder' : 'intermediate links' : intermediate cases, actual or hypothetical. actual or invented, sharpen our eyes to formal connections (GB 133), whch need not be by way of common properties - as is evident in the case of family-resemblance concepts' : 259-60.
The most useful point here is the reference to GB = Remarks on Frazer's Golden Bough in Wittgenstein : Philosophical Occasions 1912-1951 : 133, where LW refers to 'the understanding which consists precisely in the fact that we "see the connections" [Zusammerenhänge]. Hence the importance of finding connecting links [Zwischengliedern].
He elaborates :
But an hypothetical connecting link should in this case do nothing but direct the attention to the similarity, the relatedness of the facts. As one might illustrate an internal relation of a circle to an ellipse by gradually converting an ellipse into a circle; but not in order to assert that a certain ellipse actually, historically, had originated from a circle (evolutionary hypothesis), but only to sharpen our eye for a formal connection.
This is hardly crystal clear but the 'connecting links' are, presumably (I speak tentatively), the related positions an ellipse passes through in the process of being converted into a circle. Without these hypothesising or conceiving these, one would not 'see any connection' between an ellipse and a circle.
I wish I could be more definite but at least you might find it useful to check out The Golden Bough, if you've not already done so. The GB passage does throw some light on the gnomic PI, §122.