No, interleaving might account for a 600kbps or so, nothing like the difference you are seeing.

As long as you switch off the modems before mucking about with the sockets you won't confuse the exchange. Perhaps the bell wire (if attached) is the cause of one lines poor performance.

Mmmhhh...

Have checked both sockets and the face plates are not connected to the phone line as in the linked diagram. My face plates are connected using a plug on the back of the faceplate into the test socket. I cant see any way to disconnect a ring wire without getting further into the box. Am I doing something wrong?

Coincidently, since restarting the router the sync speed has dropped from yesterdays 3,692 to a meager 2,866. Have tried the test socket and it is the same.

Suppose the positive is that it might drop so low that BT have no choice but to admit there is a problem

If connected the ring wire will be attached to the detachable front part of the faceplate as in diagram 4 here: http://www.dslzoneuk.net/socket.php?type=html

Thanks John, in that case there is definitely no ring wire attached. Shame I was hoping I would be able to get the fast line to go faster

have you tried plugging (just) a phone in to see if there is any noise?,
then test again with phone, modem and filters etc.

Just tried a phone and i cant notice anything unusual in the sound quality however this was in quite a noisy server room. No scratching or hiss. Testing with the phone attached doesnt seem to have improved the speed...

In reply to a post by clueless2:

Is there information on the web that lists the key stats I have posted and what poor\acceptable\good numbers would look like?

Try http://be.nfshost.com and put your stats in there.

Several points:

Interleaving and fastpath can account for the diffference in latency.

As for the attenuation figures and snr. The lines will go from you to a local cabinet. From there they may be in different cables back to the exchange. Typically, 50, 100 or 200 lines in one cable and they are large. Two cables can take slightly different routes and hence a different attenuation level. When they reach the exchange they may be connected to different frames and from there to different DSLAMs which may be on different floors. All of those can add up to a noticeable difference in line length and hence attenuation. It is also a fact that depending on where in a bundle your line is, the SNR level can be different and there may be different interfering factors - for example atmospheric noise is more likely to affect lines on the outside rather than the inner, but outer lines will pick up less cross talk.

nice explanation

also copper thickness can vary so one bundle may have better copper than another bundle.