Upon reading Stephane's memo the nagging question I have is what good does it
do to calibrate to 1% if one is missing a larger percentage of flux than that
 in one's maps.  It seems to me that this spec also needs a corollary
spec to go along with it, that to calibrate to gamma percent accuracy in
some band also means to obtain better than (1 - gamma) percent of the flux.
That is, that the calibration is with respect to the true sky brightness
distribution on all spatial scales.  That is a very tall order.

I think it doubtful this can be achieved without the ACA.

Jack Welch just called with his own calibration strategy which he will
present at the meeting.  He notes that the Sloan survey plans to reach
1% accuracy on 10^8 objects.  This will be a standard against which ALMA
will be held.  So it would behoove us not to relax the calibration goal just
because we cannot get to within (1 - gamma) percent of the flux.
Sloan uses a small telescope which observes stars in the large
survey telescope's field.  The data taken with this telescope are accurately
calibrated and that calibration is transferred to the objects seen with the
large telescope.

Jack thinks ALMA could employ such a scheme profitably.  A first blush sketch:

One of the ACA antennas could be used for this purpose.  It would have a ring
of horns, one for each band, which would be switched in to measure a planet
in total horn power for comparison with the ACA antenna total power and the
interferometric power observed with the ACA and ALMA.  The horn experiment has
been done at BIMA at 1mm with good results; this provides the standard
flux measurement which is transferred to the calibration antenna by observing
with it and the ALMA standard feed system.  This would be done with an ACA
antenna, as the small antenna's larger field of view is an asset in the submm.
Then that antenna is a standard,
and its calibration can be transferred to the others and to the set of
QSOs which will be used for e.g. phase calibrators on that day.  Each day
or so one runs through this set of observations to give the array its
flux calibration standard.  This way, the flux calibration is built into the
phase calibration.  The planets will of course need to be stable; Uranus is
known well and as we discussed tis week Serabyn and Pardo's work at CSO hold
hope for obtaining the accuracy we need with others.  Whether this scheme
achieves 1% accuracy must be assessed.

Can we point the ACA?  Vertex has been asked if the optical guiding camera
can be mounted stiffly enough to the antenna structure to follow the radio
pointing to within tolerance.  At least one antenna manufacturer has suggested
that this could be done but the answer from Vertex is still pending.  With
a 4" telescope one can reach 12m with S/N~15 in a second, good enough to get
to 1/30 beam pointing.  Such stars are easily found in the night sky and
would be useful for offset guiding e.g. with mosaicking also.  In some nearby
dark clouds there may be a problem but for most of the sky probably not.
What about daytime?  S/N ~ N/sqrtN where the denominator has photons from star
plus atmosphere so one would need to integrate down.  Experiments are underway
at BIMA on this; details need to be investigated such as flat fielding, 
filling up the CCD wells, etc.  But results so far suggest one can get to
10m in the daytime, which means a star within one half degree or so.  

In this way one can achieve 1% flux or so (needs to be investigated; memo
forthcoming but probably not before the meeting) and very accurate pointing;
the ACA field of view is essential to operation of the scheme.

Let's discuss this at the meeting on Monday.