Interesting - I find even with sub arc minute polar alignment accuracy to do 20 - 40 minute shots at 2.3 metre focal length my guiding has to achieve better that arc second per hour accuracy on the guiding side to give really tight stars.
The maths is well thought out and presented in this article. Real world I find if you want really tight stars on long exposures at long focal lengths - well your tracking and guiding requirements are really tight.
By combining 20 - 40 minute subs of M16 and M20 last night, and knowing my image scale and pixel size - I can estimate drift to be around 0.5 - 1.5 arc seconds per hour in my shots (guided) - which seems phenomonal to me!
