Does anybody know what kind of light bulbs was used in streetlights until they were replaced by MV or fluorescent lamps, or in areas where discharge lamps were unsuitable or too expensive?
Since regular incandescent lamp lifetime is 1,000 hrs. it'd mean relamping every 83 days with avg. 12 hrs./daily use.
In case of rough service 5,000 hrs. lamps, it's about 13 months.
But it's still much less than 4- or 6-year HID lamps typically used these days.

Did they use some special lamps, transformers producing a bit lower voltage for streetlight power supply or did they just relamp as madmen?

Real incandescent streetlighting used rough service long life lamps. Lantern / lamp and its lifetime :

http://streetlightonline.co.uk/eslabulbgone.htm



Incandescent conversions of Mercury lanterns - esp. the 80W and 125W sizes (E27 base, widely available lamps...) were pretty common in private property here (Israel) in the past

Actually that aren't really conversions. The common 60s/70s Mercury lanterns are remote gear design. The user would get the lantern alone without the gear box (sold separately), and install it as incandescent

I reckon seeing numerous of those (years ago) in the motel village Ramot Shapira in Beit Meir. See at the far corner of the roof here :

https://upload.wikimedia.org/wikipedia/commons/2/26/Ramot_shapira.jpg



The incandescent on Mercury chokes used to happen in East Germany....

http://www.lighting-gallery.net/gallery/displayimage.php?album=2222&pos=0&pid=60476
http://www.lighting-gallery.net/gallery/displayimage.php?album=772&pos=0&pid=81392

The ballast output voltage is dependent on the current drawn through it. At low loads (low power incandescent) the voltage would be near the ballast OCV, the higher load, the lower voltage

On 230V (Europe) :
Ballast OCV is 230V, the incndescents are for 230V

If the incandescent lamp is low power (40W..60W), it won't be very different from using it normally. The higher the lamp power, the less voltage it will get, so the more it will be dimmed by the ballast. At some point, you just can't get any more light as if you use higher power lamp it will just be dimmer

You won't be able to get anywhere near the correct light output, no matter what incandescet you use

Lamp life will be longer, esp for the high wattage lamps (as they are underpowered the most)

On autotransformer ballasts (USA) :
Ballast OCV assume on the order of ~300V, assume incandescents rated for 120V

Low power incandescents will get high voltage, so i guess white light for a time between few seconds and few hours. At some point there may be a lamp that loads the ballast just sufficiently to get its output down to 120V, then itll work as any lamp on 120V. Beyond that point it'll be same as in Europe



When the lamp burns out an arc starts in the bulb. Here it is ballasted by a choke, and won't reach the current it takes to blow the stem fuses

On SON / MH gear it will end up in something melting or exploding and a vacuum loss. On Mercury i am not sure - It is possible the same will happen (if the arc stays), and it is possible that the current will be too low to maintain the arc, then it will extinguish on its own (as there is no ignitor to force it to keep arcing)



At very high wattages you might risk drawing too high current through the Mercury ballast and overheting it

It will happen at lamp wattage that is somewhat higher than the ballast current rating. The underpowered lamp takes less current than at full power, but not proportionally to the voltage since the filament resistance is lower when it is underpowered

Assume 80W Mercury lamp at ~0.8A current. 200W 230V incandescent takes ~0.87A. On the ballast that would be down to maybe 0.6 (guessing), so its fine. 300W lamp takes ~1.3A, now that could end up drawing more than 0.8 through the ballast

I've seen HPS fixtures relamped with incandescents before, with ballast still connected (USA, 120v 60Hz).  Sometimes they're really dim, but I wonder if the ignitor will indeed cause an arc when the bulb does finally burn out?

I've also seen CFLs in there, in one fixture in particular, and it seemed to work perfectly fine for many years, then got a high wattage incandescent, which is much dimmer than it should be!

On a 100w ballast, the filament breakage has to be small and a gap of a few mm's to src, and the arcs aren't strong. If you put a incandescent into a mh of respective wattage, the higher voltage from the ballast will overdrive and shorten lamp life, and maybe a bigger and stronger arc.

To the original question:
The bulb replacements of each few months was just common, just nobody thought about it as too short, in the contrary, 100 years ago it was assumed as rather luxuriously long time...
Don't forget the arc lamp rods had the life of just one or two days, Yablochkoff's candles only one night, the gas lanterns had required the lamp man to light each of them (the pressure wave ignition systems came later), so an incandescent bulb with life of three months was already a tremendous improvement.
And don't forget the early discharges didn't had the rated life much longer, 1500 hour was rather common for fluorescents, the medium pressure MV's had rated life of about 3000hours, so the same as the long life incandescents already available at the time.

@Medved: Yeah, I didn't realise there really wasn't an alternative with better lifespan at that time. Also manpower used to be cheaper than today so no problem with employing more electricians-relampers.
I can imagine times (and it might not be so distant future) when streetlights will be relamped (or rather replaced in a plug-in style) by drones.

@Ash: Thanks for interesting information and links.

I can imagine times (and it might not be so distant future) when streetlights will be relamped (or rather replaced in a plug-in style) by drones.

Not that much the drones, but just treated as e.g. the guard rails: Sometimes someone smashes them, so they need repair, sometimes some fault would require a repair, if inferior quality materials would be used, it deteriorate and need replacement in a short time and so on.

That allready is how you maintain proper HID lighting....

Good HPS lamps (ie. before Philips and Osram started selling Yaming lamps) can go unmaintained for ~10 years before they start failing in mass, and being HPS, they maintain well the light output levels. When the 10 years are up you have to do maintenance anyway eg. cleaning, repainting (the column) - sometimes..... Could as well do that along with repairing/replacing other outdoor stuff like guard rails, trash cans etc

Here in the UK I'm sure I heard that they simply used 260 volt bulbs in streetlights so on a nominal 240 volt supply they lasted nicely. On other mains voltages say 210 volt they used 220 volt lamps you get the idea

   According to the McGraw- Hill book," Street Lighting Practice" (1930),the life of standard "multiple" incandescent street light lamps(120volt) was 1000 hours. The life of "series" incandescent street light lamps was 1350 hours. They have a sample street light maintenance contract that would be negotiated between a town and utility to provide and maintain series street lighting fixtures. Under a 4000 hour lighting schedule( from 15 minutes after sunset to 30 minutes before sunrise)it would be required that every bulb be replaced"within 10 days of April first,September first and December twentieth". It was far more economical to "group relamp" than to have trucks going out every night to "spot relamp" burned out bulbs.
By the 1950's,all the large lamp makers listed series bulbs of 2000 and 3000 hours,and 120 volt multiple lamps of 1500 and 3000 hours. This allowed group relamping intervals of three or two times per year. By the 1960's 6000 hour lamps were listed that allowed a once per year relamp.
 But these longer relamping intervals came at a price. For example, if you were using a 10,000 lumen lamp on your major avenue,you could install a 575 watt bulb that you replaced three times a year,or a 620 watt bulb that you replaced twice a year,or a 690 watt lamp that you replaced annually. You had to balance the cost of the higher electricity against the labor of lamp replacement.
   In the late 1950's,New York City had about 200,000 incandescent street lamps and were group relamping twice a year. That meant electricians were going up and down ladders 400,000 times a year! ( yes ladders,no bucket trucks 'till 1962!) And that does not count the trips for early burnouts or fixture cleaning!
  When the conversion to mercury vapor began in earnest in NYC in 1961, the group relamping period went to once every three years,a tremendous savings in labor.

 Forgot to mention that in the mid 1970's,most of the lamp makers were offering krypton filled incandescent multiple lamps in the 1000 lumen(105watt) and 2500 lumen (205watt) with a 12,000 hour rated life. This gave a little better than a two year relamp interval. Worth it if the fixtures were in the boondocks and hard to get to.

With group relamping schedule I doubt there were special trips fro lantern cleaning, I would rather guess all the cleaning, inspection and eventually some fixes were done at once with the relamping...

Incandescent lamps were also made for street series operation. They made special lamps for 5.5A, 6.6A, 7.5A, 15A, and 20A circuits, 6.6A being the most common. They actually have low voltage filaments, like for example a 2500 lumen street series lamp was 24V and 6.6A. Such circuits were regulated by constant current series regulators which adjusted the voltage to properly operate the number of working lamps on the loop.

What happened when one lamp on a series loop burnt out ? Did they have a mechanism similar to Christmas lights to shunt out the bad lamp ?

Yep. If i understand correctly it was a thin ceramic(?) part holding springy contacts apart. At open circuit voltage it would break down, shatter and let the contacts close

Wonder how reliable it is if some sort of vibration or earthquake makes filaments in few lamps break at the same time during the day, and then the voltage is applied to the circuit with few of the isolators in series. Will the leakages to Earth present along the circuit make for sufficiently inequal voltage drop so that there is a 1st one to go, or they will stay like that without lighting ?