I have been trying different things trying to make the image fit the screen. The voltages that Sams list for pin 1 should be 155-245V and for pin 6 170-270V, the range that I measure is 216-260 on both pins. it is very little difference on the width between the two positions for the width control. I checked all the resistors and caps, changed the ones out of spec. The only thing that is different from the schematics is the width control, i measure 10K and the schematics says 20K, but it looks original.
The screen shot was taken between two screens that is why looks weird.
How can i solve this problem?

The first and easiest thing is to try a different 12AU7 tube. The 12AU7 is pretty wimpy for the task, so any weakness could be your problem. The next thing I would look at is the B+ voltage. You have selenium rectifiers, and unless you replaced them with diodes, your B+ might be a little low. Also, the .0047 coupling capacitors should be replaced regardless of any tests.
====
Ron

Sorry i rushed the post to watch the game. The B+ is at 318. I tried a few 12AU7 and there is one change, as i turn the width control i loose horizontal sync, as i change the horizontal hold, the image is duplicated. I have a gremlin somewhere else.

It looks to me that the schematic contains an error: C92 (or is it C82) should connect to pin 6 of the 12AU7. The circuit makes no sense as you have shown. I checked the
Rider schematic (book 6-9)/

Riders also show the B+ at 360v.
=====
Ron

One thing has been tripping some up is what the source is.
Run the set with NO signal, just snow. If the whole screen lights
its a source problem. If it still has bad width its the TV.
Best source is a VCR or early DVD & modulator. May not work perfect but
good enuf to fill the screen.
At that point you have to narrow it down to the stage level so as not to
waste time, $$, or risk making matters worse.

Multiple images may be normal. Later sets with coil hoz hold can
get it if turned too far. Tin whiskers in a pot can too.

73 Zeno
LFOD !

Yes I picked up the that error earlier, I thought that whom ever worked on the set before made a mistake, but as you said the circuit doesn't make sense.
I was out for the weekend, i'm returning to this in the next few days.

Understood, i'm using a sencore VG91 and VTA92 to inject the signal, i can varied the size of the signal, very precisely.