Now Win7 loaded on iMac is nice.

The development tree split at windows NT 3.1 versus win95.
The DOS tree was dos, then win 1/ 2 / 3 / 3.1 / 3.11 / 95 / 98 / Me. While windows 95 and up did a better job of hiding it, they all many many DOS related elements hidden in their kernels.
The NT kernel started with joint work with IBM on OS/2, then jumped to NT 3.1 / 3.5 / 3.51 / 4.0. NT started with 3.1 probably because Windows was up to 3.1 and started a new Windows family at 1 would have caused much confusion.
At version 5, then stopped calling the modern branch of the tree NT and just called it Windows 2000 (server or professional). The kernel for Windows 2000 for server and workstation were identical (I diffed the actual binary files one day – I was bored and curious). The differences were in registry settings, helper programs, etc.
At this point, the underlying architecture had 5.0 pasted all over it. If you can get to a Windows 2000 box, start a command prompt and see what it says. But at Microsoft HQ, it was a quarter (25 cents of money for those not familiar with US currency) in the penalty jar for pizza at the end of the week if anyone called it 5 instead of 2000.

Windows XP was 5.1. The kernel changed a good bit, but it was not a complete rewrite. They added the Windows 98 look and feel, and improved Plug-n-Play capabilities. But it shared no code with the 95/98/Me product tree. It was a significant, but not a major, upgrade to Windows 2000.
Vista was then version 6 (complete rewrite of the kernel). It took too long (5 years from XP to Vista) and had many problems that have been discussed endlessly elsewhere.
Windows 7 is another amost complete rewrite of the kernel. The confusing thing is that the command prompt (or a program-to-OS version request) will return 6.1.
This was done so that software installers wouldn't complain about it being the “wrong” version and refuse to install. Older software installers will simply think windows 7 is an update to Vista and continue to install. They did this in response to complaints that occurred when Vista was released where software that would *run* on Vista would not *install* on Vista because the installer (wrongly, IMO) looked for a 5 in the major version number and would not install if it didn't find it. This explains the comments Nash about application compatibility.
my two cents.

The development tree split at windows NT 3.1 versus win95.
The DOS tree was dos, then win 1/ 2 / 3 / 3.1 / 3.11 / 95 / 98 / Me. While windows 95 and up did a better job of hiding it, they all many many DOS related elements hidden in their kernels.
The NT kernel started with joint work with IBM on OS/2, then jumped to NT 3.1 / 3.5 / 3.51 / 4.0. NT started with 3.1 probably because Windows was up to 3.1 and started a new Windows family at 1 would have caused much confusion.
At version 5, then stopped calling the modern branch of the tree NT and just called it Windows 2000 (server or professional). The kernel for Windows 2000 for server and workstation were identical (I diffed the actual binary files one day – I was bored and curious). The differences were in registry settings, helper programs, etc.
At this point, the underlying architecture had 5.0 pasted all over it. If you can get to a Windows 2000 box, start a command prompt and see what it says. But at Microsoft HQ, it was a quarter (25 cents of money for those not familiar with US currency) in the penalty jar for pizza at the end of the week if anyone called it 5 instead of 2000.

Windows XP was 5.1. The kernel changed a good bit, but it was not a complete rewrite. They added the Windows 98 look and feel, and improved Plug-n-Play capabilities. But it shared no code with the 95/98/Me product tree. It was a significant, but not a major, upgrade to Windows 2000.
Vista was then version 6 (complete rewrite of the kernel). It took too long (5 years from XP to Vista) and had many problems that have been discussed endlessly elsewhere.
Windows 7 is another amost complete rewrite of the kernel. The confusing thing is that the command prompt (or a program-to-OS version request) will return 6.1.
This was done so that software installers wouldn't complain about it being the “wrong” version and refuse to install. Older software installers will simply think windows 7 is an update to Vista and continue to install. They did this in response to complaints that occurred when Vista was released where software that would *run* on Vista would not *install* on Vista because the installer (wrongly, IMO) looked for a 5 in the major version number and would not install if it didn't find it. This explains the comments Nash about application compatibility.
my two cents.

English is not Martin's first language, and I guess that might be the same for other websites you read.

How's your German?

Regards

EB

Windows 3.1 and Windows NT 4.51 were not of the same architecture use or design. If you want to group anything it should be with Windows 95 and Windows NT 4. Then 98 and Me possibly with 2000. XP, where the versions between consumer and commercial-business were first merged should be stand-alone and not grouped with 2000, they were different Operating Systems, albeit with few major modifications, but enough that it shouldn't be considered a .5 or minor release.