As Andreas Blass and others surmised, this is indeed an abstract. It was for a 10-minute talk at the Annual Meeting in January, 1975. Here is the entire "paper":

**On Manifolds with nonnegative Ricci curvature II**

Let $M$ be an $n$-dimensional
Riemannian manifold with nonnegative
Ricci curvature. Then the exponential
mapping $\exp_p$ for any $p\in M$ ,
restricted to the domain bounded by
the cut locus, is everywhere volume
decreasing.From this fact one deduces
the following THEOREM. Let $M$ be a
Riemannian, $n$-dimensional, complete
manifold with nonnegative Ricci
curvature. Then, if $r$ denotes the
injectivity radius and $D$ the
geodesic diameter of $M$ , the volume
$V$ of $M$ satisfies $V \ge c_nr^{n-1}D$, where $c_n$ is a positive
constant depending on $n$ . In
particular, if $M$ is not compact (i.
e. if $D=\infty$), the volume of $M$ ,
under the same assumptions, is
infinite. (Received November 6, 1974.)

A couple of notes. I reproduced the capitalization and (non)hyphenation of the title as it appeared in the *Notices*. I also tried to preserve some oddities in punctuation in the text, but otherwise "TeX-ified" it; the original is literally typed, with a handwritten $\in$ symbol.

**Added 11/13/12:** Out of idle curiosity, I went back to the library today, to see if Calabi ever gave a talk titled "On Manifolds with nonnegative Ricci curvature I." If he did, it wasn't at an AMS meeting (or else I didn't dig back far enough).

In the process, however, I noticed that, beginning in October, 1972, the *Notices* ran a "Queries" column, inviting "questions from members regarding mathematical matters such as details of, or references to, vaguely remembered theorems, sources of exposition of folk theorems, or the state of current knowledge concerning published conjectures" -- i.e., a sort of snail-mail version of MathOverflow. Here's the inaugural query (the answer to which arrived in a then-speedy three months):

1 . R.P. Boas (2440 Simpson Street, Evanston, Illinois 60201). Given a
finite collection of vectors, of total
length 1, in a plane, we can always
arrange them in a polygon, starting
from 0, that at some stage gets at
least $1/\pi$ away from 0. Mitrinovic
[*Analytic inequalities*, 1970, pp.
331-332] cites Bourbaki [1955], but
the theorem was known at least in the
early 1940's, when I remember seeing a
paper on it; can anybody supply the
reference?

My own favorite is from the next issue:

4 . Cleve B. Moler (Department of Mathematics, University of New Mexico,
Albuquerque, New Mexico 87106). Can
somebody recommend a good source where
I can learn about the connection of
mathematics and various biological
processes such as photosynthesis?

Cleve Moler is perhaps best known as the inventor of MATLAB.

In conclusion, the answer to the OP's question, "is the *Notices* available before 1995?" the answer seems to be yes, but only at libraries that hold onto old journals. I wonder if the AMS could be persuaded to make the early volumes of the *Notices* available through JSTOR.

Noticeshas evolved from being just a formal record of meetings (with abstracts), elections, etc. Among the abstracts published were quite a few that didn't correspond to talks actually given at meetings and that may or may not show up in published work later on. At that time theBulletinpublished more detailed multi-page research announcements, but these too didn't always have follow-up. – Jim Humphreys Nov 13 '12 at 0:46