< Actually air flow is quite complicated. Using muffin type fans <makes it worse. Here are some basic rules of thumb. Yep! <1. Fast moving air in an open space will find a surface < and run along it. Pressure boundary. <2. The only way to make uniform air flow is with restriction. Also a must as you cannot control freely moving air. <3. Rotating air will do strange things until straightened < out. ( examples, air coming from a rotary fan and also < most air going through a single small restriction ). Vanes unrotate air. <4. Laminar flow is best understood but turbulent < flow removes more heat. The trick is to get turbulent air that goes where you want it. < I have seen the above problems cause all kinds of effects <that were not obvious at first. The best case I know is the 8xxx and 9xxx series VAXes as they were air cooled and also tried to be quiet and efficient in cooling the contents. Both series would cook in minutes if the cooling failed (or if the phase rotation was wrong!). The Altair cooling was poor at best even with a 125CFM fan (noisy too). the IMSAI was a bit better, the compupros s110 rack better still. The best s100 crate was the Intergrand and TEI boxes as they clearly were built to address the heat problem. An example of how rough this is the NS* horizon I have has a 115cfm fan with a filter pushing air in and the box has varios strips of manilla (144pound stock) to block and route the air or the heat build up on the teltek HD controller and the z80 master processors (z80, 64k sram, 2sio and ttl mapping and bus master logic) would make them shut down even with a modified PS. Allison