Over the top motion of the antenna may offer several benefits but by far the most useful is
that it offers a simple and effective way to measure the offset of the elevation and azimuth axes
of the telescope.  This offset acts like a baseline error and therefore may affect ALMA astronomical
imaging in the interferometric mode.  Over the top motion may increase the antenna cost, however,
so a consideration of the ability to measure the offset in other ways should enter a decision on
whether to include over the top motion as a specification on production antennas. 

At IRAM, observations to determine the baseline are made in good conditions (and short baselines
to minimize atmospheric phase terms).  Observations over a wide range of elevations and dozens of
sources are made.  The phases phi_ij are fit for baseline i and antenna j as:

phi_ij = C_i + X_i*cos(H_j) + Y_i*sin(H_j) + Z_i*sin(D_j) + E_i*cos(El_j)

 X_i, Y_i, Z_I (equatorial coordinates of the baseline errors), and axes offset E_i are
computed to minimize the weighted sum of square phase residuals, where H_j, D_j, El_j are 
respectively the hour angle, declination and elevation of measurement number j.

E_i is measured for the baseline, not the individual antenna, but it is relevant to baselines.
At SMA, which does not go over the top, the value of E_i is about 0.1mm or less for most antennas.
While it is not necessary for the antenna to go over the top to measure the offset, it makes the 
measurement much much easier--measure phase, go over the top, measure again and the axis offset of 
an atenna is determined.  Without over the top, one needs as much range of elevation as possible.

An error analysis of the solution above needs to be performed for an ALMA implementation.

For ALMA, the baseline errors must be less than 65 microns ('Calibration of ALMA' document).
The errors in determination of the offset of the axes must then be much less than this, and the
offset should remain stable in time.  The degree of stability and the length of time are issues to
address.

The baseline plan with ALMA is that from the most compact configuration, four atennas are moved
every four days.  This continues to the largest configuration and back again until the antennas return
to their original location in the tightest configuration.  At present this cycle takes several months.
so the stability of the measurement of the axis offset should be such that offset errors do not contribute
to baseline errors over this time period.  Conservatively, the axis offset should not vary by more
than 20 microns over nine months, if the antennas do not go 'over the top' and hence the offset is
not easily measured during the period of antennas being on long baselines.  During this period, each
antenna may be moved many times.

If the 'over-the-top' motion is allowed, each antenna's axis offset will be measured when the new
baseline is measured; on average this will be about every two weeks.  In this model, the stability
might be somewhat less, small enough not to affect the baseline measurement, but over a period of
two weeks.

Antenna IPT should investigate whether a metrology solution is practical.