Thats what I thought it should be but the legend has overwritten itself. Matched pairs never really remain matched though in reality so I would expect some serious distortion to creep in.
Should be fine enough for this application. Ideally they'd be on the same bit of silicon and actually matched like the now dead and gone LM394 or inside an opamp. But the only thing you're going to have an issue here with is DC offset between the input dividers. And meh to that. This would be an insignificant contributor to the distortion in this application. Saturating the transistors is a risk here as the output range is only about 5V with this circuit
![Scared :scared:](https://www.eevblog.com/forum/Smileys/default/scared.gif.pagespeed.ce.oZptHNcbsD.gif)
If we take the whole thing to bits it probably makes more sense...
The core of it is a fairly lame opamp:
![](https://imgur.com/9l3Duhi.jpg)
And this is just surrounded with a feedback loop to give it some gain and some coupling to make it single supply:
![](https://imgur.com/CNCcZEV.jpg)
So going back to the original question, Q4 is the input transistor on the non-inverting input in the lamest opamp that the human race has produced
![Laughing :-DD](https://www.eevblog.com/forum/Smileys/default/smiley_laughing.gif.pagespeed.ce._hfWAz_QHO.gif)
Edit: I haven't measured THD formally but it's around 0.44% from the simulation assuming you don't saturate the output transistors which is pretty good for a shite design
![Smiley :)](https://www.eevblog.com/forum/Smileys/default/xsmiley.gif.pagespeed.ic.R8GFI-pF6f.png)
Edit 2: improvements I could make: Current mirror in diff amp, better VCCS, JFET diff amp. But can't be arsed.
Aaaand back to less interesting things
![Sad :(](https://www.eevblog.com/forum/Smileys/default/xsad.gif.pagespeed.ic.L3FGyzQrjB.png)