The Jaguar is another aircraft that has a reasonably well documented inertial autorotation mode. It is a mode that exists at low angles of attack whereby the principal inertial axis is depressed below the flightpath vector. Essentially, autorotation occurs in this situation when the rolling inertia exceeds the aerodynamic roll damping and thus the roll continues with no lateral control applied. Note that this is a completely different mode to aerodynamic autorotation which occurs at a high angle of attack and which may result in a spin.
Multiple rolls can trigger inertial autorotation as the roll rate increase progressively when the inertial axis diverges from the flightpath vector, thus increasing the roll inertia. However, a far more role relateable trigger in the Jaguar is entering a roll at less than +1g, or simultaneously pushing and rolling such as when rolling out of a hard turn or trying to reverse the turn direction as quickly as possible without climbing. The minimum normal acceleration for entering a rapid roll in a Jaguar is +0.5g with no external stores (and then only for a maximum bank angle change of 180 deg), and +1g with external stores. It is worth considering next time you watch the RAF Jaguar display that a vertical roll is actually a 0g rapid roll i.e not a permitted manoeuvre in a Jaguar!
It is possible to see the onset of inertial autorotation in the Jaguar even during 360 deg rolls entered at +1g. As the roll progresses, the indicated normal acceleration reduces to around 0. At high speeds of 450 KIAS and greater (low AoA) this occurs after only a 90 deg bank change. However, at lower speeds (higher AoA), the bank change before this is observed increases until at 250 KIAS (minimum speed for rapid rolling) it does not occur at all.
The recovery from inertial autorotation in the Jaguar is to smoothly pull the stick rearwards until the motion stops (whilst maintaing neutral lateral and directional control inputs). This agrees with the theory that to stop the motion you have to reduce the angle between the inertial axis and the flightpath vector and so should, theoretically, work for any aircraft so long as its longitudinal controls are still effective. A very interesting topic!