It appears to me that the arrow is simply indicating signal flow as right to left in that particular wire, which is different from the remaining circuit's left to right. You would do well to try and locate the "unknown source" that this schematic was "borrowed" from.
That website is well known. In this circuit I see many problems. I don't have time to find all of them, so here is a sample.
1) The most critical component is the coil L1. The text says 15 turns and the schematic says 20 turns. I suppose either will work, but VC1 will be different. I would try 15 turns first.
2) VC1 is 0-22pF. A 0pF minimum is unnecessary, and it would be a difficult to find part. With a 15 turn coil, the nominal setting for VC1 is about 18pF, so a 5-15pF part in parallel with a 5 or 10pF fixed capacitor will be easier to find and easier to adjust.
3) The 4 inch coil and 5.5MHz operating frequency will give a very limited detection distance. Might be good for finding nails in the wall, behind the plaster.
4) The oscillator V+ should be regulated and isolated from the audio amplifier. Otherwise zero beat will interact with the audio, since the audio amplifier can draw a few hundred mA.
5) R2 value (330R) is wrong. It would prevent T1 from ever conducting. Maybe it's 330K.
6) C11 and C13 are basically in parallel. VR1/C11/C13 low pass frequency varies from infinity to 5kHz with the volume setting.
7) The power bypass capacitors C1 and C14 are unlikely to be effective at the 5.5MHz operating frequency. The detection oscillator will tend to lock on the ceramic filter's frequency. The circuit needs a few 0.1uF (100n) caps to filter the oscillator V+.