In other words, no matter what the frequency of VCO2 is, it will be restarted by the frequency of VCO1. Same thing for VCO3 independently of VCO2.
It's a fixed configuration, so VCO1 could never be a slave, and is always the master.
That's why the "sync effect" will always work best when VCO1's frequency is lower than that of the slaves. Moreover, the "sync effect" will only be heard when changing the frequency of the slaves manually, since there is no envelope, or LFO (VCO3 used in an LFO configuration) routing to the pitch of either VCO2 or VCO3 independently of VCO1.
BTW, for those with a newer osc board (the one with more than 7 adjustment trimpots), I don't know of the sync mod shown on Kevin's Synthfool page for it would work. As Kevin himself wrote a disclaimer that he did not try them out himself. I just know that the one for the "regular" osc board works fine in my Mini.
Top one in the picture below is the most common, below that is the "newer" version.

Actually it might look more like this one (note the ua726 round chips instead of the DIP package of a CA3046 with a Tempco glued on top, as the ua726 as already a integrated thermostat system inside it)
But you get the idea that it has more than 7 trimpots also.