I must tell you that you're doing great book keeping on the name.  I have similar info/details on Hoover. 

But, excellent work on your part!

book on cleaning?

I got a useful bit of information from a former Eureka (1978-2014) employee:

So, that's a little background. Now, to your questions. Model numbers were the main identification control, of course. But we would have run out of model numbers in no time if we hadn't used the model letter suffix, or "type", as it was labelled on the data plate, to indicate changes that affected servicing the machine. So, your 1428 might have been a 1428, type A. We might then have changed it to a type B if a part changed that was not interchangeable with mating parts. Or, I remember when we changed the handles from steel to ABS plastic we advanced the letter on models that were affected. Model numbers were indeed used to control which retailers got which models. We used higher specifications (amps, height control, type of bag, etc.) for department stores and vac shops (they were major players back then), and retailers like KMart, Catalog stores, etc. had different model numbers with different specs. There was no quality difference in how the models were manufactured. But specification differences in convenience features, filtration efficiency, design trim, and of course motor types. We private-labelled for Montgomery Ward and JC Penny, so they used their own model numbers.

As for serial numbers, I'd have to say I'm taking an educated guess. When I started with the company we bought pre-printed data plates so those most likely were grouped by model number. But eventually we produced our own plates, and then eventually produced them on each manufacturing line, as I recall. The serial numbers started with the week and year of manufacture. But the numbering system changed in later years, and I can't recall in what way. In general, a series of cleaner was built on a dedicated production line, so it is likely that you would have a long string of sequential numbers on the same series, but individual models might only have 500 or a 1000 run at a time before changing to a different model in that series. And then of course, the next week the numbers would start over with the new week.