like everything today, capitalism carries on today. the only generator that is resistant to deviation is the seesaw used by Veronika. it is very expensive to produce.. big and clumsy to make, but if you want better than others, this is one solution..

NRG FOUNDER A swing that has not been beaten to this day. NRG (veronika).
Stephen J Moss started to swing 10 years before this article was published. https://dn790002.ca.archive.org/0/items/RadCom_Magazine_1989-06/RadCom_Magazine_1989-06.pdf
For this brilliant invention he is grateful to Rene Mesni 1874-1949  inventor of the kallitron power oscillator.
"
Symmetrical oscillator
While testing VHF radio links, René Mesny encountered a lack of frequency stability in the transmitters he used. In 1924, he designed and developed an oscillator circuit with two triodes mounted in a symmetrical push-pull device.
"
#
A kallitron power oscillator/transmitter that made history as AC4YN, providing the first amateur contacts with tibet in the 1930s.  derived from T urner's oscillator as then presented  (reproduced in 7T. September 1988). He draws attention to a form of two-valve oscillator developed  by Mesny in the early 1920s from which an output of several watts at 150MHz could be obtained: no mean achievement at that time. G8FEQ writes: "In the early 1920s there was a great deal of interest in the VHF bands despite the poor performance at high frequencies of the components of the day. One of the workers in this field was Mesny. His circuit (Fig 2) shows that this is a push-pull oscillator resembling the 'pseudo-kallitron' but with the unusual feature that feed-back Is by inductive coupling (as used in the traditional regenerative detector)"  Neither G8FEQ nor I can trace anything further on the work of Mesny. but it is worth pointing out that both the Franklin and Ross Gunn oscillators of the 1920s (both included in Theory and Design  of Valve Oscillators by Dr H A Thomas, 1939) use cross-coupled two-device configurations basically akin to the G40TC form of kallitron oscillator, although in the case of the Franklin the tank circuit is connected to earth rather than between the anodes. A solid-state kallitron also turns up in Sprat (Issue Nr 38) built by John Beech, G8SEQ to provide a test signal source on 70MHz: GSSEQ notes that in the form shown it can be continuously tuned, frequency modulated, or scanned by applying a sawtooth waveform to the varicap diodes. It was built using copper PCB pads stuck to a groundplane of plain PCB material, 30mm square, the layout resembling the schematic.
#
https://groups.google.com/g/alt.radio.pirate/c/S26ZH3TMsJM
https://www.gareth.net.nz/hollings/Radionecks%20%E2%80%A2%20Pirate%20Radio%20Talk%20%E2%80%A2%20AAREFF%20(page1).htm
https://www.gareth.net.nz/hollings/Radionecks%20%E2%80%A2%20Pirate%20Radio%20Talk%20%E2%80%A2%20AAREFF%20(page2).htm
https://www.gareth.net.nz/hollings/Radionecks%20%E2%80%A2%20Pirate%20Radio%20Talk%20%E2%80%A2%20AAREFF%20(page3).htm
https://www.gareth.net.nz/hollings/Radionecks%20%E2%80%A2%20Pirate%20Radio%20Talk%20%E2%80%A2%20AAREFF%20(page4).htm
VERONICA FM AT WAR! please read
Kallitron oscilator  by Stephen J Moss bring to us:
The improvements, the real advantages, The oscillator to tune the entire band without the need for trimmer adjustments and so that the voltage change required to deviate from 75 kHz is large enough to make the modulation sensitivity sensitive. If you only need a few mV for full deviation, the equipment will be prone to hum and noise problems.
A big improvement was the addition of another pair of varicap diodes for audio quality.