The simulator in LTSpice looked OK ... so I moved to the breadboard with an LED (eventually replaced the LED with a buzzer) at pin 3. One more difference in the assembled circuit is the condenser at pin 5 is not electrolytic.
The supply is a retired AT SMPS from my old computer.
The waveform in LT Spice is initially LOW, and subsequently HIGH after a few hundred seconds. The assembled prototype however begins with the buzzer audible.
My impression is, therefore, that cycle has somehow got inverted in assembly. Is this impression correct? What need be done to have the output on the prototype behave as designed?