Because HMI had better colour rendering, efficacy, life, consistency and optical performance

Although the first six foot tube was introduced by GE of America in 1944 as part of its Slimline T8 series, it remained a rather uncommon size.

The first preheat cathode six foot tube was introduced by British Lighting Industries (a division of Thorn Lighting) in 1966 under the Atlas Super Six name.  It had a T12 diameter and was rated 85 Watts.  BLI introduced it as new standard with highly optimised performance of the total lighting system, and it was originally intended that this size should become the dominant tube for the British fluorescent market.  Previously the highest runners in Britain had been the 5ft 80W tube, and after 1956 the 8ft 125W.  Both were highly loaded tubes that offered impressively high light output for industrial and commercial applications, but that made them rather inefficient - their loading being similar to the American HO (high output) tubes introduced by GE in 1952.  That situation was partially overcome in 1961 when BLI introduced the new standard of the 8ft 85W tube.  That delivered an extremely impressive boost in lighting efficiency - but as noted by RRK such long tubes were considered unwieldy and difficult to handle.

The compromise of 1966 was to introduce the 6ft 85W tube as an intermediate model that satisfied the following criteria:
- high light output suitable for industrial and commercial applications
- high efficacy to attain lowest cost of operational ownership
- the maximum length tube that could strike easily on the British 240V mains supply, without expensive or complex control gear such as autotransformer ballasts, or the capacitive ballast as used for the 8ft 125W tube

The 6ft 85W indeed quickly became one of the most popular sizes in Britain but curiously it did not enjoy the same success in other countries.  As RRK also already pointed out, for low voltage countries in the Americas it would not be an advantage, because the high tube voltage would require a greater open circuit voltage of the autotransformer ballasts, resulting in greater size, weight, cost, and electrical losses of the ballasts.  In other European countries it was also not adopted because their mains voltage of 220V was slightly too low to ensure reliable starting of the six foot tube, which was not a problem on the British 240V mains.  As such, the six foot tube standard remained almost exclusively limited to the UK and other 240-250-260V countries around the world.  Another restriction was that in continental Europe Osram and Philips had recently constructed new high speed tube production machinery - but not having foreseen the six foot development and also not having ever made the 8ft tubes in significant volumes, they built their lines to handle a maximum tube length of 5ft.  It was therefore commercially very difficult for them to follow the Thorn lead in the new 6ft business.

Two years later in 1968, Thorn made another major development with the introduction of its "Superwhite" phosphor.  That delivered an almost unprecedented boost of 6% in luminous efficacy, and allowed the 6ft tubes to rival even the 8ft 85W in total system efficacy.  The same phosphor was not attractive to apply on 8ft tubes because those already delivered enough light, and the high cost of the new material would have made them too expensive on a longer tube.  Also in 1968 Thorn developed a remarkably efficient new semi-resonant-starting ballast for the 6ft 85W tube, which eliminated flicker on startup and greatly extended lamp life.  These two achievements further cemented the 6ft 85W as the leading British tube.

Following the global Energy Crisis of the early 1970s, in 1973-74 Thorn re-rated the 6ft T12 from 85W to a dual-rated 75/85W tube.  New ballasts were introduced to run the tube at slightly lower current, which caused a drop of 10W in power consumption but due to the lower power loading, the decrease in light output was much less significant.  Thereafter, most new 6ft installations used the newer 75W ballasts.

Finally, following Thorn's 1975 introduction of the first Krypton-filled T12 energy-saving tube in Europe, and especially after Philips' 1978 extension of the Krypton technology to the new T8 formats as T12 retrofits (first in 4ft 36W, then 2ft 18W and 5ft 58W), it was a logical step for Thorn to apply the same principle to its 6ft tubes.  That resulted in the introduction of the 6ft T8 70W krypton lamp.  Like its 6ft T12 predecessors, it remained almost exclusively used in countries having mains voltages of 240V or higher.

Stainberg street and other streets at Kiryat Benjamin neighborhood, used to have AEG Triangel lanterns with Eltam gear and 150W HPS lamps. One of the lanterns that I could see through my window at my room at my hostel, had Eltam ES-PI 1000 anti-cycle ignitor. I used to see them restriking at power outages, but they were replaced by 3000K LED lanterns, and the only visible HPS lanterns remained are the two AEG Triangel at the path between Stainberg and Yalag streets. And they are the only lanterns that extingushes during power outages. The LEDs simply blinks like incandescent during power outage.

They're not gone yet here, but i do miss them a bit on the highways.

The golden yellow light of HPS always separated 'Away from home' from 'Home' for me. As soon as the lights change from HPS to fluorescent, you know you're in a residential area, which i find a pleasant switch from a travel mindset to a being home/relaxation mindset.

I have few memories about LPS lit streets. In 2021 i thought for the first time 'Let's make a picture of the LPS streets' because in 2020 we still had one big inner city artery lit with LPS, but literally just months before i wanted to capture those scenes, they turned out to have been removed.

The good thing is that i now have all of those lamps myself and could get them for cheap because for the professional user, the lamps are now practically worthless.

Looks like most CMH lamps have smaller electrodes than quartz MH lamps, as most of them are rated for electronic ballasts.

My 50w and 70w SON/CDM/CDO drivers are both electronic.
My LPS ballasts are inductive, as are my 150w HPS, 100w SDW-T and 50/80w MV ballasts. I tend to call those ballast, and an actual switching/oscillator circuit a driver because it's an active thing rather than something just being a impedance in series with sometimes an ignitor that only serves to start the discharge.

But interesting to know how those things work and what makes them different.

Aside from the difference in power factor, extremely important is also the current crest factor which determines the required size of the electrodes.

Magnetically ballasted lamps suffer a high CCF which causes considerable electrode heating cycling variation during the 50/60Hz sinus, and the electrodes must be made large enough to accommodate that.  These significant heat losses reduce lamp efficacy and shorten the life.

At higher frequencies above about 400Hz, or on squarewave operation, the current peaks are drastically reduced.  The electrode size can be reduced drastically - just look how small are the CPO electrodes vs the equivalent SON types.  Even if you can set up a magnetic ballast that operates the lamp at the same power as electronic (ignoring voltage and current differences which will always be significant due to the different waveform), it will quickly destroy the electrodes by severely overheating them at the peak of each current cycle and causing higher continuous temperature operation.  For this reason, electronically ballasted lamps of almost any type should not be run on magnetic gear if there is a desire to achieve anywhere close to the rated life.  In some extreme cases life may be reduced to only a few hundreds of hours.

@James: Here is the lamp: https://www.lighting-gallery.net/gallery/displayimage.php?pos=-246513

Because the particular halides do not have high enough vapour pressure to attain the desired spectral performance at the available arc tube wall temperature.  So more halide is added to raise the total vapour pressure, and improve performance.

It is maybe a rather old lamp, over time all manufacturers optimised their thermal designs to be able to reduce the salt dose.  And in Sylvania’s case changed from dosing with pure mercury to 3% cadmium in mercury to improve colour rendering without need for halides.

The bromides are more chemically active than iodides and help to protect the quartz from attack by recombining with corrosive metal vapours further away from the wall.  They are also more active in accelerating the halogen cycle to keep the wall clean from sputtered tungsten in lamps that are sufficiently highly loaded - but also have the disadvantage of being more active in causing beavering of the electrode shank at the cold spot where it passes through the arc tube wall, which can reduce life.

The bromides also have different vapour pressures than iodides, and depending on the metal may be more or less effective in bringing their vapour into the plasma.

Many metal halide lamps contain a finely balanced mix of bromides, iodides, and in highly loaded lamps also the chlorides, to balance each of these processes for optimum initial performance, lumen and colour maintenance, and life.  With some elements it is not easy to obtain the bromides, or it is desired to have a bromine-iodine mixture.  In this case they are added as iodides, and part of the metallic mercury mix is eliminated and dosed as mercuric bromide to get that halogen into the fill via an easier mechanism.  The same is done for lamps where iodine is consumed too rapidly during life, by replacing part of the mercury fill with mercury iodide.  Those are often easy to spot thanks to the intense reddish-orange colour of that salt when cold.