The night actually started out clear, but with the moon up, we
started with specoscopy: Cart 1 plate 627.  Right after opening the building,
we went to move the telescope up and found the MCP unresponsive - we could
not even connect to it via telnet. We rebooted it and all was well.

Today, Dan is driving on his machine while Scot is operating Landru in
multi-screen mode, but with a DIFFERENT Xserver than previously.  Earlier
in the evening, he got the old Xserver to die within 10 minutes of logging in.
The new Xserver runs a different instance of the window manager in each
screen and does not allow for dragging of windows across screens. But it
does mean any window that pops up from another window will be on the same
screen as the parent.  We'll run with this and see if the X crashes go away.
(Most of the time, btw, the "crashes" means the Xsession dies and we are 
returned to the Xdm login prompt - although more severe crashes have also
been experienced.

Cart 1 - plate 627 required a 15" E offset to find guide stars.  We ware
running with the 72 point pointing model. Dan recommends running with this
model and applying a 15" E offset if you can't find stars.  Last night, the
same plate with the same pointing model found the stars right away. What'sup
with that?  We did reboot the MCP today, though.

Combined with last night's data, plate 627 was finished in two exposures and
we had about an  hour to go before the moon went down. Since it was still
clear (and seeing still excellent), we decided to use this time to switch
instruments so the moon wold be down by the time the imager were ready. If we
beat the clouds, we could always do an astrometric calibration scan.