My understand was that IFR airplanes always needed a clearance from ATC. Am I mistaken about that?
In the beginning, all airspace was what is now called Class G. Some airspace was later designated as Controlled Airspace in order to give ATC the authority to control IFR traffic in that airspace so that separation services could be provided. IFR separation won't work unless all airplanes are required to participate.
ATC clearance is only required to fly IFR in controlled airspace. That's why it's called "controlled" airspace.
Us civilian pilots typically learn to fly VFR first then IFR later. This often gives us a skewed view on the purpose of controlled airspace.
I presume IFR flights are allowed to fly in the clouds in class G, like they do elsewhere? In that case, is separation between IFR flights in class G essentially based on the big sky theory?
What is IFR, the Instrument Flight Rules?
They are listed in 91.167 through 91.199. (The Visual Flight Rules are 91.151 - 91.165)
Look through the instrument flight rules for the rule
(91.173) which requires an ATC clearance and see to what it applies.
As the ATC system developed, controlled airspace was added around airports with extensions to protect approaches, and along airways. This was long before VORs.
A/N ranges and NDBs were used to define, and navigate the airways and approaches. This system allowed an IFR flight to depart, climb, fly enroute, descend, fly an approach, and land in controlled airspace on an IFR clearance. At the time, pre-WWII, there wasn't RADAR. Everything was non-RADAR separation depending on position reports, airways, and enroute holding, to keep airplane separated.
Today's new instrument pilot in the US sees a system that appears to be based on RADAR. RADAR is just the shortcut that has been added on top of the existing system which was based on NAVAIDs and airways. Non-RADAR separation is rarely seen in the CONUS today, expect for departures and arrivals at remote airports without an approach control facility or in the overnight hours when it is closed. Most of the rest of the globe is non-RADAR, though, and non-RADAR separation is still used extensively. All of the flights from the northeast US to the Caribbean are non-RADAR, giving position reports to ARINC on HF (and transitioning to satellite CPDLC).
I mentioned flying DC8s into an airport where the floor of controlled airspace is FL245. That is Diego Garcia (FJDG) in the Indian Ocean. We'd be westbound from Singapore on UR212 with Brisbane Radio (non-RADAR). They'd descend us to FL240 reaching their western FIR boundary (GUDUG), 155NM from the airport, at which point they'd terminate us. We'd call Diego Tower, which provides IFR traffic advisories to participating aircraft (there usually aren't any others). They'd read us the weather and ask us what approach we were going to fly (no approach clearance since it's uncontrolled airspace). Based on the approach, they'd tell us a fix to report then clear us to land when we reported it. Departing would be similar. We'd get our oceanic clearance, starting at the FIR boundary, the depart. Diego Tower could send a departure message, with our departure time and estimate for the FIR boundary fix, to the next facility, Brisbane going east or Mumbai going north). We'd initially climb to FL240, to remain in the uncontrolled airspace, and would start trying to raise the next controlling agency on HF. When we established contact we'd request a climb after passing the FIR boundary.
For those interested in how the ATC (should) work, and any instrument rated pilot or student, I recommend the AvWeb
Say Again? articles by Don Brown. There's a lot of good information in his articles for VFR pilots, too.
Don Brown
www.avweb.com