Short answer: if your existing Light is 300, and you wish to avoid going below 250, the three new weights could be pegged to 285, 265 and 250. Or something like that.
I do recommend not going below 250, else you will have problems in a significant portion of Windows GDI apps. If you and your client don't care about them, no worries. But my general experience is, such a decision is likely to bite you eventually.
Then again, CSS only accepts multiples of 100 for specifying weightclass for the web, despite my protests, because the W3C wants to make it impossible for us to make a font family that works everywhere. (OK, that's my slightly biased/frustrated interpretation of the discussions. Obviously I was not happy with the outcome.)
Richard, I wouldn't be in any way shocked if a future court contradicted the findings of Adobe v SSi, and Meshworks v Toyota certainly appears to be a binding precedent in the 10th District.
But your phrasing "unassailable precedent supporting the notion that fonts are not copyrightable" is a stretch, and more importantly "there have been no successful cases that have come to an opposing conclusion" is simply false. What's your argument here on Adobe v SSi? It was unsuccessful? The summary judgment did not represent an opposing conclusion?