The direct coupling of the two stages is to be avoided as the small amount of noise on the collector of the first transistor is being directly coupled to the oscillator transistor. The oscillator stage should be self-biased and the base should be held rigid as it is a common-base stage.
If that was correct, which it is not, then how would the audio modulate the RF carrier.???
BTW: its a common emitter stage, not common base.
The 22n on the power rails is effective at 100MHz and it improves the output of the RF stage enormously. An electrolytic has a high impedance at 100MHz.
What you should said, is to ADD a 22nF to the power rail decoupling, leaving the 47uF in place.!