I've been speaking with ThinkPad enthusiasts recently. This is widely regarded as a bad idea, but at least I haven't yet joined any new forums, right?
One of the things that has come up, especially in the wake of the "retro" ThinkPad project at Lenovo is what properties really define a machine, series, or product. In the case of the ThinkPad, it's a matter of finding what machines really define the product as a whole. The ThinkPad 701c is probably one of the best examples, but most people agree on various examples through history, such as the ThinkPad 600 series, or even different machines that together represent what everything was about.
Ultimately, the definitive machine for a platform or for an era on a platform needs to represent what the values and leadership of the company were at the time, ideally, what the bulk of the systems sold really were, what the target market and customers were like, and so on.
With that in mind, I would absolutely argue that the Macintosh Performa 5200 should be considered a definitive Macintosh. While I don't think that any hard numbers exist (outside of Apple) for how many of each model were sold, it's nearly a given that in 1995 and 1996, Apple was selling more 5200s and 6200s than any other model. Possibly more than all of the other models combined. In addition to being one of the most popular Macs, the 5200 and 6200 are really representative of what was happening in Apple at the time.
"But Cory, the 5200 is a total mess!" you say. Yes, and so was Apple. Apple couldn't decide what it wanted to be, and was troubled badly by a series of failed initiatives to replace the existing Macintosh System Software, in addition to problems Apple was having with policies, product line bloat, policies that disallowed Apple from selling anything with less than a 50% profit margin, which encouraged everything to cost a lot more than it was worth (and a lot more than IBM and some of the workstation vendors would have charged) or extreme cost-cutting measures.
Objectively, without thinking about its position in the world or its price, the Power Macintosh 5200 is a bad machine. I frequently defend it because of what it was and when it was built and what it did for its customers, but it was such a bad design, and it could only have come from a company that was trying to build a $1500 machine for $700 or less, in 1995 when $700 didn't really buy a whole lot in terms of computer parts.
One of the things that makes the 5200 as it actually happened important to Apple is that increasing the price would likely have reduced the number of people who could buy it, making the model even less cost effective and likely to assist keeping Apple afloat. Using a different platform was likely out of the question either for performance (the 601 was ostensibly slower than the 603) and cost (the 601 may well have cost more than the 603) reasons. One option was for Apple to just keep selling the machines they were selling, but they were trying to push the platform forward, in a time when the need to increase performance in order to run everyday applications and new things like multimedia encyclopedias and mid 1990s education games.
The successors to the 5200 and 6200 over the next few years got a lot better from a technical perspective. The Power Macintosh 6360 (and 5400/6400, 5500/6500) featured a completely re-designed board using faster everything at a similarly good price. The systems were better, but the product line was still a mess.
In 1998, when Apple introduced the iMac, their product line had shrank considerably, from about 17 models to five, with configuration options replacing models as the way people chose how much they paid for a system in a particular form factor. The iMac represented a revitalized Apple with Steve Jobs at the helm. The company was Internet-positive, building more focused computers, and fewer of them, to meet the needs of their customers better. The iMac even managed to fairly significantly under-cut the cost of its predecessor, the Power Macintosh 5400/225. (One of at least six models on sale at the time, all based on the same motherboard) which was available for $2000. The iMac would have outperformed this model, but may have been slightly slower than the 6500/300, although I don't have performance data on hand. For the first year or two, the iMac only came in one configuration. Even when colors were first introduced, you could only buy it in one configuration from Apple, which must have made it a lot easier for them to sell.
There are a lot of definitive Macs along the way. Ultimately, relatively few of them are the really high end machines that a lot of people seem to like. That type of machine is definitive for ThinkPads and PS/2s, as well as UNIX workstations and minicomputers, but the Mac was never any of those things. It was always a cross between a home accounting and letter-writing appliance, and an information and communications terminal.
I don't think that collecting high end Macs is bad (although it is a little weird) but writing the 5200 off as being valueless to the platform and the history of the platform, or a well-rounded Mac collection, is a bad idea.