A class A amplifier is defined as one which is biased to a point where plate or drain current flows for the entire swing of the input signal between the cut off and saturation. This is done by biasing the stage around the middle of its course between cut off and saturation points. More exactly the stage should be biased to dissipate the max power with no input signal, but sometime to centre the linearity of the transfer curve the bias point is shifted from the max dissipation point. The JFA has six single ended Jfet stages working in this way. The sixth stage has no gain; it simply acts as phase inverter of the signal already available at PIN 8. .

Why does JFA generate even harmonics?

It is known that tubes and Jfets use the same working principle. In the tube a negative voltage interferes with the flow of electrons running in the vacuum from the cathode to the plate, thus modulating the current flow in the tube. In the Jfet the same negative voltage interferes with the flow of electrons running in the silicon from the source to the drain, thus modulating the current flow in the Jfet. Tubes and Jfet exhibit the same smoothed shape transfer characteristic curves.  Transistors, on the other hand, have an abrupt transfer characteristic curve with a threshold knee around 0.7 V.  Tubes and Jfet produce even harmonics if used in a single ended stage, while in a push pull stage the even harmonics are cancelled.
 

What do we mean for class A and pure class A?

Once a single ended stage has to be biased two modes are available; self bias using a cathode/source resistor, or fixed bias, using a negative bias voltage and grounding the cathode or the source. The first way is the most widespread while the second is especially used in guitar power amplifiers where a negative feedback is often applied from the output to the input stage. The JFA uses self biased stages using a particular Rdrain/Rsource ratio. This guarantees a stable gain & bias and low distortion. This is what we mean by class A and it should serve to satisfy all HI FI guys, but could leave some music guy less happy, so:
the Sa, Sb pins allow the user to bypass the source resistor with a high value capacitor (100 mF suggested). This allows the stages to work purely in A, in other words the output current / drain voltage is exactly the consequence of the Jfet transfer characteristic curve. This frees all the even distortion of the stages. Furthermore the first stage of the JFA can be fixed biased applying a negative voltage at INPUT pin 2 (-1.7 Volt suggested) and grounding the relative source Sa. This is what we mean by pure class A

Some notes about operational amplifier.
Long time before the birth of digital computers, operational amplifiers born as analogical computational blocks for the first analogical computers, as the blocks were connected in order to calculate a mathematic function. Their duty at the beginning was to produce an output voltage depending on several input analogical variables and they were intended to be used with DC. However the OP developments lead to an increase of the operative speed, so someone found it useful to use the OP to amplify audio signal.  Due to the internal circuit diagram an OP is a network of multiple AB push pull transistor stages, which are in no way related to a linear amplifier, but because of their very high gain, a heavy feedback can be used to produce good results. In any case the gain of any OP decreases with the increase of frequency, thus any OP based amplifier will exhibit a higher distortion at high frequency.  The JFA can amplify a signal without using feedback because it is linear, therefore its response is the same at any audio frequency. Considering that the JFA input noise is spectacularly low (2.5 nV/Hz at 1000 Hz) this increases its ability to amplify very small signals, resulting in a fast, clean and crystalline sound.

Map