In the late 80's and early 90's, Smalley and others made claims that the C60 fullerene bearing icosahedral symmetry was the most spherical molecule known, and perhaps the most spherical that could exist. While it makes sense to me that the smallest member of the 60n^2 isocahedral family of buckyballs would be the most spherical (I believe larger members should tend towards and isocahedron?), are there really no other more spherical geometries that are possible or actually experimentally realized?
I recently took a look at: "Production and isolation of an ellipsoidal C-80 fullerene" by Chun-Ru Wang et. al. (http://pubs.rsc.org/en/content/articlelanding/2000/cc/b000387p/unauth), and it looks like C80 forms an ellipsoid. However, I'm not sure about C80's endohedral counterparts.
Update - To be clearer about what I mean by "more spherical", I'm looking for a family of buckyballs that converges quickly on the volume of a circumscribing sphere as a function of the number of carbon atoms (or alternatively, as a function of the 'k'th smallest member of its symmetry family). For example, from this Mathematica document: ("http://reference.wolfram.com/legacy/v5_2/Demos/Notebooks/BuckyballConstruction.html") we can see that the icosahedral C60 fills ~87% the volume of its circumscribing sphere. If C-80 filled a larger volume of its circumscribing sphere than C60, as the k=1 member of its family of buckyballs, I would consider that "more spherical" than C60. The problem with that assumption, however, is that C60 has 20 fewer carbon atoms than C80.
Update 2 - Another way to measure how "spherical" a buckyball is could involve looking at the distribution of angles between any two adjacent carbon bonds. One would particularly want to look at the global minimum angle(s), being careful to avoid problems with large planar faces that approximate graphene and washout outliers. If the global minimum angle between carbon bonds tended to ~(120 degrees - epsilon), as a function of the kth minimum member of a particular symmetry group, that family of buckyballs would presumably also converge on their circumscribing spheres as a function of the number of carbon atoms.