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.