null*, 1, 2, 3, null* => no gaps

where null* is any number of nulls and not just a single null at the beginning and end. Your original code makes more sense to me now.

public boolean hasGaps(List values) {
if (values.size() < 3) return false;
boolean rv = false;
List sub = values.subList(1, values.size()-1);
while(! rv && sub.size() > 0) {
rv = (sub.get(0) == null);
sub = sub.subList(1, sub.size());
}
return rv;
}

Similar in Clojure:
(defn has-gaps [v]
(cond (< (count v) 3) nil
:else (some (fn[x] (nil? x)) (subvec (vec v) 1 (dec (count v))))))

First I was puzzled about ->>, it's unknown in the release clojure 1.0.0, but in snapshot clojure-1.0.0-2009-10-28 it's working:

user=> (->> 1 (+ 3) (+ 4))
8

Woudn't it be more readable to use empty? false? instead of seq boolean. I would create a more general trim utils function like (if it's not already in the contrib lib):

(defn trim [s t]
(->>
s
(drop-while #(= t %))
reverse
(drop-while #(= t %))
reverse))

and use it like:

(defn has-nil-gaps? [s]
(some nil? (trim s nil)))

#(not? (nil? %))