There was years ago a device that monitored voltage and used an auto transformer to maintain a lower voltage yet not allow it to drop too far, in the days of fluorescent lights having wire wound ballasts it likely did save some power, however it had a lower than 100 amp rating so had to switch off with a high load, and they did not last long before switch mode power supplies resulted in them being redundant, seem to remember made in Cheshire near the Welsh border but marketed in the South of England.
I did do some experiments myself where we were required to light some tunnels on building of Sizewell 'B' and my boss decided better to use 58 watt 110 volt fluorescent, a quick calculation approx 0.5 amp each, 16 amp supply so should run 32 but to be on safe side made 25 unit strings of lights.
On fitting it was not long and they were it seemed switched off, but then used the clamp on ammeter and found using over 20 amp not 16 amp, so unit on the bench and tests, and it used around 0.75 amp, however although claimed to be 110 volt actually there was a auto transformer lifting volts back up to 230 volt, and the input was 110 - 0 - 127 so easy enough to swap to 127 volt instead of 110 volt, the first 20 in string was OK but last few would fail to strike as last 5 were set to 110 volt input. On testing was down to 15 amp so yes the adjustment of voltage had saved 5 amp at 110 volt so 550 watt.
So yes it did save power, however today we are not permitted to fit wire wound ballasts any more, so that saving is no longer the case, a 58 watt fluorescent tube today draws around 55 watt, and the output with an electronic ballast is also 10% higher in lumen, plus the high frequency gets rid of the strobe effect with rotating machinery. So a fluorescent tube lamp was considered as around 75 - 85 lumen per watt, and is now around 95 lumen per watt so in the main the same as LED input to output. However since the ballast and tube sold separate often the lumen rating and watts rating is lower and higher than really the case.
As to boiling a kettle, energy radiators from the walls of the kettle, so the slower it boils the more power it needs, so a 230 volt kettle powered from 230 volt uses less energy to boil than the same 230 volt kettle plugged into a 110 volt supply, in theory with no losses it should take around 4 times longer, in practice it takes around 8 times longer do to losses, and yes I tried it, what I wanted was distilled water to top up batteries, so did not want it to boil too rapid as needed to condense the water again, using a tea urn, normally they were boiling within 1/2 an hour to hour after a refill, mine would take nearly 8 hours before producing distilled water again.
See this report Wikipedia I will guess some has raked up a 20th Century report and republished it without checking, which shows how little point there is reading the papers.