Explaining how to install Linux in an article this size is comparable to building the pyramids using a tweezer and a toothpick. But, I can at least cover a few trouble spots and offer some cheap advice. The first document you ought to read, however, is the Linux Installation and Getting Started Guide by Matt Welsh available at http://sunsite.unc.edu/pub/doc/LDP/Installation-HOWTO.html.

Most newer distributions of Linux install rather painlessly, but there are still occasional problems with weird hardware. The type of hardware we're talking about here is basic hardware, like disk drives and scsi controllers and CD-ROMs. These devices are necessary for the installation program to even get started, normally.

When you first make your ``boot and root'' disks, you normally pick a pre-compiled kernel designed for your hardware. With Slackware you could find a kernel that would support your basic hardware configuration, install that kernel on your bootdisk, and then go from there, hopefully. Newer installation routines, such as RedHat Software's, starts a program that asks you basic questions about your CD-ROM and whether you have any PCMCIA devices or SCSI devices, and then it probes your system to make sure it can find what you say you have. In other words, the installation routine has some intelligence. It can tell you if it's having trouble, and you know what the problem is immediately. With older installation routines, you were not always so lucky.

One of the first things you should do is make sure your computer has components listed in the Linux Hardware Compatability HOWTO in the Linux Documentation Project (http://sunsite.unc.edu/pub/Linux/docs/LDP). Of course, you'll need to check to see how recently the document was revised, since new devices are constantly being added. If you have a very new machine with recent devices, e-mail the author of the HOWTO with questions (fric@pobox.com).

Most hardware related problems arise with very new machines. Since it's very rare that device manufacturers build drivers for Linux, it's necessary for device drivers to be built by the Linux community. This they do very well. Still, if you are using some oddball SCSI card or some unusual chipset from Mars, you may run into problems. In the beginning, some Cyrix CPUs had problems with the kernel, simply because most Linux users had been using Intel CPUs and Cyrix support lagged behind Intel. However, I hear there's plenty of Cyrix support out there now.

Your best bet is to stay with mainstream hardware devices that are about six months to a year old. Also, stay away from ``Plug-and-Pray'' devices in general. You'll probably just have to disable that feature in most cases anyway. And then verify that your hardware is specified in Linux Hardware Compatability HOWTO.

Many installation problems in the beginning seem to be related to X Windows, the graphical interface most commonly found in a Linux system. At this point, this is more a configuration problem than an installation problem, strictly speaking. But more than likely, you'll very much want the X-Windows graphical environment on your system.

X11R6 is the technical name for X-Windows. You may obtain a free version of it under the GNU public license (XFree86) or you may purchase a commercial version. XFree86 is what normally comes with a Linux distribution, although the ``official'' Red Hat Linux comes with the Metro-X commercial X server. Some other commercial distributions include something similar. The advantage to commercial X servers is that they include support for more diverse hardware.

We use the term X server to denote the program that ``serves'' X clients their graphical environment information, such as basic paths to fonts and other system resources. When you install your X server, you must choose one that supports your video card and monitor. For example under XFree86, if you use a Diamond video card with an S3 chipset, you would use the S3 X server that comes with XFree86. If you were using the Metro-X commercial X server, you would configure it with its own tools. Each X server is a little different, and you must read your documentation carefully.

Under X Windows, you must also select a window manager. This tool handles the actual appearance of the GUI and the behavior of its objects. One of the most popular is FVWM, a free window manager that builds upon TWM, a slightly older window manager. This would be the program that tells your applications how to minimize and maximize, where to place a clock on your desktop, or how many virtual desktops to maintain. Many other window managers exist, some making your desktop look like NextStep, or OpenLook, or even Windows 95.

XFree86 has a reputation for being rather challenging to configure. But once configured, it is robust and fast. The Linux XFree86 HOWTO by Matt Welsh goes into detail about how to configure XFree86. It's available as part of the Linux Documentation Project. Configuring FVWM requires reading the man page for it. Fortunately, the fvwmrc file is well documented and explains even more clearly than the man page.

Sometimes circumstances dictate unusual installation procedures. Perhaps you don't have a CD-ROM, or maybe you must install from a network server. Linux can even be installed via ftp over the Internet, although this is not recommended unless you have a dedicated T1 connection or better! Most distributions are much too big for floppy disks nowadays, however, so you might just as well rule that method out.

These methods are all possible, and even others. If you must resort to alternative methods of installation, I recommend you carefully read the documents mentioned above and consult one of the Linux newsgroups or mailing lists dedicated to installation.