The Irish Bomfords

Geodetic Surveys in India 1930‑35

Guy Bomford

This is some description of the geodetic work of the Survey of India during the period 1930‑35, in the first years of the Survey Review. It is in the form of personal reminiscences, and in detail refers only to such of the work as I was personally concerned with.

The 1930‑31 period

Summary. At the end of 1930 I was at Dehra Dun, the H.Q. of the Geodetic Branch, in charge of the Computing and Tidal Party, with Latitude and Magnetic detachments in the field. E A Glennie was in charge of 14 Party (Pendulums) working in Rajputana. I M Cadell was in charge of 15 Party (Triangulation), working in eastern Burma, with a detachment (M N A Hashmie) in the extreme south of Burma. No 17 Party (levelling), with H P D Morton in charge, was working as usual with three detachments on high precision levelling and five on lower order work. The Director of the Branch was J de Graaff‑Hunter.

Triangulation: Eastern Burma. The programme of Cadell's own detachment was to observe at seven stations to close a primary triangulation circuit, including a connection with the Siamese (now Thai) system, and then to measure a baseline at Kenton (21º15' N, 99º30' E), with its extension to the adjacent primary. From central Burma a motor road went as far as the Salween, and in fair weather at the right time of the year, motors could reach Kengtung. Elsewhere there were plenty of tracks fit for mules and ponies, but these animals had to come from Yunnan. The base level of the country is below 1000 feet in the Salween and Mekong valleys, but elsewhere hill tops are between 5000 and 8000 feet.

By mid‑December, working with a 5½‑inch Wild theodolite, Cadell had completed the Siamese connection and had closed the primary circuit except for one station, when he began to get ill with pneumonia. He managed to reach his last station on the 18 December, three days march from the main road, but died there ten days later. Cadell was very tough. He travelled light and took little notice of feeling ill, but it is hard to see what he or anyone else could have done about it.

News of Cadell's death reach Dehra Dun on the 31 December, and on the 4 January I left to tidy things up, and to complete the programme if possible. From Dehra Dun it took nine days to reach Salween, where mules were waiting for me, and I got to Cadell's last station on 17 January. The long delay had naturally disarranged the helio squads and it was another 12 days before all were in place and the observations completed. After that measurement of the baseline and its extension went smoothly.

Kengtung baseline. This was the first baseline measured in India since 1882. At Kengtung there is a flat open plain of rice‑fields about ten miles by five with hills all round. an ideal site for a baseline. Cadell had reconnoitred the site the previous year, and an assistant L R Howard had spent November and December 1930 clearing the line. building stations, and laying out pegs for the tripods every 24 metres. For the base measurement we used six invar wires, two simultaneously for the south to north measure and two others for north to south. The other two were used as sub‑standards against which the working wires were compared every morning. I had standardised all the wires in our Dehra Dun observatory in August 1930, and they were of course re‑standardised on our return in April 193 1. For their coefficients of expansion we had planned to rely on their Sèvres (1908) certificates, but as soon as we started work it was apparent that these certificates were wrong, as was shown by the difference between the two working wires between early morning (13ºC) and the afternoon (33ºC). To get a provisional figure we set up tripods over ten 24 m bays and made repeated measures with all the working wires in the morning and in the afternoon.

The squad of about 35 survey khalasis and a recorder* had had some preliminary practice at Dehra Dun, and after the first two days everything went very smoothly. An assistant P K Chowdhury and I read the wires, and Howard with 15 khalasis set the tripods over the pegs, and recorded their heights above the pegs. The heights of the pegs were obtained by levelling along the line before and after the measurement. To start with we only managed 60 bays a day, but later increased it to 120. When we had finished we re‑measured the first two days' work, where the discrepancies between fore and back were greater than elsewhere.

*Khalasis are members of the most junior grade in the Survey Department, and include helio men, chain men, instrument carriers etc. In the 1930s their pay was between £1 and £2 per month according to skill and length of service. A recorder or computer was a member of the Lower Subordinate service, which also included plane‑tablers, traversers and draughtsmen, the backbone of the department, whose pay was between £3 and £10 per month. Assistant, as used in this article, implies members of either the Upper Subordinate service or of the Class II Gazetted service.

Triangulation: Southern Burma. While the work described above was proceeding, Hashmie's detachment was working on a connection between the long Burma Coast triangulation series and the Siamese triangulation in the Kra Isthmus (latitude 10ºN), observing at eight stations with a 12‑inch micrometer theodolite. They had severe trouble with rain, cloud and thick jungle, and one new station which was placed on what appeared to be a good hill with a clear view to the east had to be re‑sited after observations had been made to it from the west. This practically involved a fresh start. Worse still, an outbreak of beri‑beri resulted in the death of six khalasis. Fortunately, an improvement in the weather in January and February made it possible to get the connection completed.

Summer 1931. Arriving back at Dehra Dun on 24 March, I resumed charge of the Computing and Tidal Party. In addition to the usual routine there was plenty to do. In the observatory there was the re‑calibration of the baseline wires. They were found to have held their lengths well. I also determined their temperature coefficients in the comparator, and this confirmed the figures got in the field reasonably well. There was also the computation of the triangulation, and of the work of the Latitude and Magnetic detachments, described below.

Latitude detachment. The latitude detachment under R B Mathur observed astronomical latitude at 44 stations at intervals of about ten miles along meridian 75ºbetween latitudes 33º30' and 26º03', using a small zenith telescope on two nights at each. This was intended as the beginning of an extensive enterprise, namely to cover the whole of India and Burma with a network of geoidal sections, along which the deviation of the vertical (both components except for meridional sections) was to be measured at intervals of about 15 miles. Geoid sections had previously been observed elsewhere, notably in Switzerland, but this project as completed in 1940 involved 20 000 km of section. There was to be a main east‑west section from Indo‑China (now Laos) and Siam to Persia (Iran), and a meridional section from Cape Cormorin to Kashmir, side‑stepping a little as it crossed the east‑west section. R B Mathur's work in 1930‑31 gave this project a very good start.

Magnetic detachment. Terrestrial magnetism is not a geodetic subject, but with the apparatus of 1900‑50 a Survey Department was a very suitable organisation for carrying out a magnetic survey. A full magnetic survey of India and Burma giving the declination, the horizontal and the vertical force, and their annual changes, at 1425 stations had been carried out between 1909 and 1922. To keep it up to date required observations every ten years at about 60 Repeat Stations, with continuous recording at four or five fixed observatories. This was now due to be done, and in 1930‑31 Shyam Narain, the magnetic observer at Dehra Dun, observed at 37 stations controlled by the Dehra Dun and Alibag (Bombay) observatories, the only two then functioning.

Reorganisation 1931. In 1931 financial crisis and retrenchment fell upon India and its Services. Since 1920 the geodetic parties nominally existing had been No 13 astronomical, 14 pendulum, 15 triangulation, 16 tidal, 17 levelling, 18 magnetic, 19 baseline, and the Computing office. By 1930 the Tidal party had amalgamated with the Computing office, and both the Magnetic and Baseline parties were in abeyance except for H.Q. activities, which also lay with the Computing and Tidal party. The astronomical party also had only taken to the field in 1923‑24 and 1924‑25. We were thus able to meet the demand for retrenchment by formally abolishing 13, 16, 17, 18 and 19 Parties, leaving 14 Party for Pendulum and Astronomy, 15 Party for triangulation and levelling, and the Computing and Tidal party which comprised the Computing office, the Observatory section (including standardisation of baseline wires, and the magnetic observatory), and the Tidal section which made tidal predictions for 40 ports between Suez and Mergui. This reorganisation did us little damage because we had been training assistants who could carry on the work in charge of detachments on considerably less pay and overheads than were required for independent parties. It had in fact been largely anticipated in the field season 1930‑31, and we survived without any significant loss of out‑turn. Our greatest loss was our Director of the Geodetic Branch, J de Graaff‑Hunter, who had to retire at the age of 51, two years before it would have been usual for him to retire on long leave pending retirement. Two topographical Director's posts had been abolished, and the Geodetic Branch had to find room for one topo Director along with a topographical Party.

Our experience may be of value to others faced with the recurrence of similar problems. A source of further saving, apparently not now possible, was that all Government servants except the most poorly paid (such as those on £1 or £2 a month) had their pay reduced by 10% during 1932, 1933 and 1934.

The 1931‑32 Period

Triangulation. In the early winter of 1931‑32 I observed a primary triangulation chain, known as the Dalbandin series, along meridian 65ºE in Western Baluchistan, connecting the Kalat and Mekran longitudinal series in latitudes 29ºand 26ºrespectively.

A railway runs from Quetta, through Nusliki, Dalbandin, Nokkondi, and Mirjawa, to Zahidan (formerly Duzdap) in Persia, but in 1932 trains did not run beyond Nokkondi. West of Nusliki there was a train each way once a week, with an extra train loaded with water about once a fortnight. Western Baluchistan is a dry and barren area, devoid of trees, except in the south near the coast, with conveniently placed hills of between 6000 and 8000 feet. At the right time of the year it is ideal country for triangulation. Water is a possible problem in places, and I always carried enough for one night, but it can generally be found in the hills, and its shortage caused no delays or inconvenience. An assistant, Khushal Khan, joined me to go ahead building new stations and placing helio men. He was a most excellent man, expert at making friends with local chiefs, and he smoothed my path. I used 13 helio squads of three men each, an exceptionally large number of squads because I planned to move fast, and apart from the squads at the first three stations, there was going to be no possibility of their overtaking me and getting into place at forward stations before they were needed. There were 16 stations in all, including two old stations at each end, and two new stations which were observed to, but not at, for a future extension to the west. We started work at Dalbandin on 1 November, and finished at the south end of the chain on 14 January. We had 55 camels, a few donkeys, 31 survey khalasis excluding the helio squads, and Khushal Khan and I each had an 'escort' of nine locals: useful for finding water, taking an odd camel to withdraw helio squads, and so on. I also had a recorder to book readings. We carried a month's supply of basic food, which we were occasionally able to replenish. The camels could find adequate grazing.

I used a 5½‑inch Wild theodolite, and carried a spare, but our Wilds (which had given some trouble in earlier years) had settled down, I had no theodolite trouble. I did not observe horizontal angles at night, because the lamps of those days were bulky, and would have much complicated the placing of helio squads. Some of the lines were very long, one of 80 miles. In clear weather in December, the helio was brightly visible at that distance, and even had to be damped down, but long rays make for refraction trouble, and are not recommended if avoidable. My average triangular error was 0.71", maximum 1.89". This is within the Indian definition of primary triangulation, but it is not of the highest quality. I blame the long lines, and the absence of any night observations.

I observed a short program of circum‑meridian latitudes at all the stations along one flank of the series and azimuth at five stations, to provide a weak geoidal section along the series.

Persian Boundary. I was back at Dehra Dun at the end of January, but another piece of work soon turned up. The boundary between Baluchistan and Persia (now Iran) runs southwards to the sea from the hill Koh‑i‑Malik Siah, the trijunction of Afghanistan, Baluchistan and Persia. In 1896 T H Holdich demarcated a central section of 100 miles between Kuhak and the Hamum‑i‑Mashkel swamp, but the hot weather intervened, and 150 miles northwards from the Hamun were left undemarcated. There were a few lines of description, defining the boundary, but no pillars or adequate map. The Governments of India and Persia had shown no inclination to complete the demarcation, but in February 1932 both parties suddenly decided that it was desirable. So at three days notice I set off for Quetta with instructions to make a map of the boundary area north of the Hamun, and to put up boundary pillars if agreement with the Persians (now Iranians) was possible. At Quetta I collected Khushal Khan again and two surveyors, all expert plane‑tablers, and after some delay we reached railhead at Nokkondi on 8 March, and Mirjawa on the frontier four days later. At this time the weather was cold and clear.

At Mirjawa I met the Persian representative, Sirtip Abdur Razak, who had served with the Turko‑Persian boundary commission in 1913. He remembered H C Cowie of the Survey of India, who had served on the same commission, and he was a ready collaborator, although hesitant French was our only common language. Unfortunately, he had no instruction to demarcate, so we could only make a map and recommendations. As he had no survey party with him, the map making involved no disputes or complications. North of Mirjawa there were a number of triangulation stations and intersected (unvisited) points at the west end of the Kalat longitudinal primary series, enough for plane‑tablers, but to the south there were none. Fortunately, there were a number of suitably placed small hills, and I observed a quick chain of triangles from Mirjawa down to the Hamun, while the plane‑tablers got busy surveying a ¼ inch to the mile map about 20 miles wide along the boundary. They worked rapidly, surveying about 350 square miles a month in the intricate hills north of Mirjawa, and 800 in the open area to the south. The map was finished by mid‑May, which was fortunate because it had then become impossibly hot, as my predecessors had found. I think it was hotter than I have ever known it elsewhere.

The map showed that the written description left over from 1896 contained ambiguities. Neither Government seemed anxious to pursue the matter, so it again lapsed. Since 1950 I think something has probably been done.

The 1932‑33 Period

Base measurement. The programme of the triangulation party in 1932‑33 was to measure three baselines and their extensions in Burma, at Mergui, Amherst, and Kalemyo (23º20' N, 94ºE) with I H R Wilson in charge, in preparation for which I standardised the six invar wires. The result was disconcerting. Since the last measurement in 1931, one 24‑metre wire had decreased by 0.77 mm, one had increased by 2.69 mm, and the other four had changed by between 0.10 and 0.28 mm, all these figures being larger than they ought to be. Clearly some extraordinary misfortune had occurred, but when or how could not be discovered. However, the wires had been restandardised, the field party was ready to depart, and there was nothing to be done but to carry on and hope for the best.

Wilson duly measured the three baselines with daily comparisons between the working wires and the two kept as field standards, which showed that there was at least reasonable stability. Restandardisation at Dehra Dun in April 1933 showed changes of at most 0.24 mm (1 in 100 000) in the four working wires, and of 0.35 and 0.42 in the field standards. These figures are larger than one would have liked, but in the circumstances they might have been worse. The six wires were then re‑measured in May 1933 to give the current rate of change of each wire. Given the lengths of the wires in September 1932, April 1933 and May 1933, and their relative lengths at the times of the measurement of each of the three baselines, I deduced a probable figure for the length of the mean wire at each baseline, which had an apparent probable error of 1 ppm or less, and I felt confident that the error did not exceed 1 in 300 000. In view of the likely error in the base extensions, an error of 3 ppm in the baseline itself is no disaster. But we had learned that damage to a wire may make it unreliable for at least the next year.

Geoid section. For the field season 1932‑33 I planned to observe a geoid section across Burma, to start the long section from Siam to Persia referred to earlier. I used a 60ºJobin prismatic astrolabe, but since the astrolabe has no impersonal device, I also look with me a portable bent Transit with a so‑called impersonal micrometer, which I used in addition to the astrolabe at every fourth station. My personal equation with the Transit was determined by four nights observations at Dehra Dun both before and after the field season. Anticipating a little, the field value of (Transit minus Astrolabe) varied between +0.08s and +0. 14s, mean +0. 12s. So in fact I got substantially the same result as I would have if I had got along without the Transit. I regretted not having calibrated myself on the astrolabe instead of on the Transit.

For my triangulated (geodetic) position I had chosen a line through country which had been fairly recently surveyed topographically, and there were plenty of recognisable topo triangulation stations and intersected points, generally hill tops. During the preceding months I had selected promising sites from the maps about one day's march apart, say 12 miles in hilly country, and had obtained the coordinates of points likely to be useful. So at each station I observed angles (with a small Wild) to six or eight recognisable points, and generally observed a Polaris azimuth at Sunset. In a few cases I traversed from a single nearby point, or set up a short base and found a distance.

The routine was that we marched one day, and pitched camp at or close to a place where I could get a triangulated fix, which I at once did. The next morning I computed the fix semi‑graphically to ensure that all was well with it, and set up the astro instruments in time to get the Rugby 16.30 (Burma time) signal. After sunset I observed for two hours with the astrolabe (30 stars), and finally got the 00.30 Rugby signal. I carried three chronometers which I compared with each other at wireless times, and during the star observations. While waiting for the 00.30 signal I marked up the chronograph sheets and worked out the comparisons between the three clocks, as a precaution against unsuspected trouble there. We moved on again the next day.

Every fourth station where the Transit was to be used, required eight time stars and two azimuth stars with the bent Transit, about two hours work after completing the astrolabe. Apart from the trouble of setting up the Transit and getting it adjusted in level and azimuth, it set me back with reading the chronograph sheets and the clock comparisons, some of which had to wait until the next station. It was in fact a considerable nuisance.

I had with me a Survey of India recorder, 20 khalasis, 15 coolies from the Wa States (north of Kengtung), about 40 ponies and mules from Junnan with a dozen Chinese drivers, and two interpreters. Burma does not produce ponies, mules or coolies, and I got them from Syed Ismail of Kengtung, who had done the same for Cadell in previous years. Syed Ismail was a very useful man, with much influence in Kengtung. He was a Chinese Moslem, of which there is a colony in Yunnan.

We started work at Monywa, on the Chindwin. The coolies and ponies arrived only 12 hours late, having travelled the last 200 miles by rail, preceded by 200 miles on foot from Kengtung, and about the same again from their own countries, ‑ Syed Ismail's work. We left Monywa on 2 November, and worked through Gangaw and Haka (22º40' N, 93º20' E) in the Chin Hills up to the Indian frontier. We then marched back to Monywa and worked eastwards through Ava, Taungyyi and Mongpan (20º20' N, 98º20' E) to the Burma‑Siam‑ Indo China (now Laos) trijunction, south of Kengtung, 44 stations in all, finishing on 24 February. I then observed longitude for two nights at Kengtung to complete a Laplace station there.

Summer 1933. Longitude. In the hot weather of 1933 preparations were made for the International Longitude project which was to be undertaken in October and November. The first reliable determination of Greenwich longitude in India had been made by S G Burrard and G P Lenox‑Conyngham in 1894‑96, which gave the longitude of Dehra Dun observatory as 5h 12m 11.77s. They measured the longitude from Greenwich in several legs, changing their positions, fore and back, for alternate legs. They used portable Transits for local time, and the electric telegraph for clock comparisons between stations. In October and November 1926 the first International Longitude project had been organised, in which about 50 observatories all over the world made nightly observations of local time, and so far as possible received the same wireless time signals. At Dehra Dun we had used two Transits with impersonal micrometers, and a 60ºJobin astrolabe. Two observers used each instrument for alternate weeks. The final result was 5h 12m 11.79s.

In 1933 three Transits were used: one with a hand driven micrometer, one motor driven, and one with a new 'moving shutter' device which had been designed by de Graaff‑Hunter, and constructed in our Dehra Dun workshop. The observers were B L Gulatee, R B Mathur, H C Banerjea and J B Mathur, each using one of the three instruments for (on average) one third of the time. The final result was 5 h 12m 11.78s, agreeing very nicely with earlier results, but the three instruments taken separately gave an unsatisfactorily large range of values: 11.68 for the hand driven Transit, 11.84 for the motor drive and 11.80 for the Hunter shutter. There was clearly 'personality' as between the instruments, much more than between the observers. It was satisfactory to have the Hunter shutter in the central position, but by 1939 de Graaff‑Hunter had not been able to get the idea taken up by instrument makers in the UK.

The 1933‑34 Period

Summary. This season I measured three more baselines, while R B Mathur and J B Mathur each observed lines of geoid section. It was a notable year for Glennie's pendulums, as he observed at no less than 71 stations, 41 in India, 21 in Ceylon, 8 in the Maldive Islands and one in the Lacadives. Levelling followed its usual routine and included some relevelling in Bihar, and near Pagu (Burma), which had recently been disturbed by earthquakes.

Baselines. The three new baselines were at Padag near Dalbandin in Baluchistan: at Poona: and at Namtiali, a small place in north‑east Assam. After the trouble with the wires in 1932‑33 I took no risks. Standardisations in April, May, July and September 1933 showed that the wires had stabilised by July, and further it was not inconvenient to visit Dehra Dun to remeasure them between each of the three baselines. Measures made in November '33, January, February and March '34 showed practically no significant changes. The length of the mean wire at each baseline was certainly known to within 1 part per million.

L R Howard and P K Chowdhury prepared the baselines for measurement, Chowdhury and I read the wires, and Howard emplaced the tripods, as at Kengtung, except that at Padag, C A Biddle, who had recently joined the Department, took Chowdhury's place at the other end of the wire. I observed the extensions at Padag and Poona, but at Namtiali M N A Hashmie, who had made a preliminary reconnaissance of the baseline, observed the extension. Climatic conditions at Padag were inconveniently variable with a maximum working temperature of 27ºC on 26 December, and a minimum of ‑6ºC on 4 January. Apart from this the weather was consistently fine.

When observing the Padag and Poona base extensions I tried to investigate horizontal refraction trouble by observing sets of three measures of all angles on each of 6 or 8 zeros at four different times of day, namely before 10.00, between 13.30 and 16.00, from 17.00 to sunset, and between one and four hours after sunset. The Padag extension had 11 conditions between 16 angles, and Poona had 21 among 30, so the difference (adjusted angle minus mean observed) for any one of the four daily periods gave a fair measure of the mean error due to refraction in each angle during the period concerned. Tentative conclusions were reached about the relative advantages of including night and/or afternoon observations at different types of station, but refraction is a tricky matter, and I would not claim that my conclusions were of very general application.

Triangulation in Assam. The Assam Valley triangulation series of 1867‑78 was a good quality secondary series in its western half between longitudes 92ºand 94ºbut it then entered flat ground and degenerated. It was therefore planned to convert it to primary standards by reobservation and realignment, and to connect it to the Burma triangulation by a new primary series through the Naga Hills. After finishing the Namtiali base extension, Hashmie repaired and cleared 24 old stations, and built six new stations in preparation for observations in 1934‑35.

Geoid sections. Bengal. R B Mathur extended my Burma section westwards as far as longitude 86º, observing at 35 stations with the astrolabe, and at every fourth station with the portable Transit in addition. It was generally possible to observe at previously fixed triangulation points. He determined his personal equation at Dehra Dun, before and after each season, for both instruments. As in the previous year, it was apparent that the Transit had added little to the accuracy of the astrolabe, and in future years much trouble was saved by relying on the astrolabe only. Transport in the Chittagong Hill Tracts was by 35 khalasis and 70 hired coolies. In the Bengal delta it was by carts, coolies, or boats, with frequent changes from one to the other. In one march of 15 miles five such changes were necessary.

Geoid sections. South India. Work on the meridional section line was begun by J B Mathur, using the small zenith telescope and observing 10 pairs of latitude stars in one night at each station. He observed at 36 stations between latitudes 8ºand 17º, at or very close to existing triangulation stations. At this time, J B Mathur was graded as a computer in the Lower Subordinate service of the Department. He was doing work which had normally been done by personnel two or three grades senior to himself, for which he deserved great credit.

The 1934‑35 Period

Summary. This year Glennie was on leave, and Hashmie took over the pendulum observations, and observed at 33 stations in Kathiawar and Rajputana. G H Ormaston observed at 30 stations of the new Assam Valley series, referred to before. Levelling carried on as usual. In addition we were able to concentrate three detachments on the geoid sections, and to complete the two main lines from Persia to Indo‑China and from Cape Cormorin to Kashmir.

Geoid section. Baluchistan. Knowing the country and its existing triangulation fairly well, I did this section myself, using the astrolabe, and generally resecting position from several rather distant fixed points and a Polaris azimuth. I had motor transport in the form of two 30 cwt. lorries loaned by the Army, and took with me six khalasis, four drivers and cleaners, and a local escort of two. As far as Nokkundi, water was obtainable from the railway, and petrol by the weekly train. The weather was perfect, and it was possible to observe 39 stations between 17 October and 2 December. At its eastern end (in longitude 70ºE) this section joins on to Glennie's line of nine latitude and longitude stations, made while he was making pendulum observations in the Indian desert in 1930‑31. At two primary triangulation stations at which azimuth had previously been observed, I did a two‑night programme instead of one, to complete Laplace stations.

Geoid section. Orissa to Rajputana. On my return to Dehra Dun, I handed the detachment over to R B Mathur, who then extended his 1933‑34 section westwards from longitude 86ºto 75º(57 stations), where it joined his meridional section of 1930‑31, and he then observed a further 11 stations joining his 1930‑31 section to the cast end of Glennie's desert line of stations. He observed his personal equation at Dehra Dun at the beginning and end of the work, and also at Fyzabad in the middle. Fyzabad is an old longitude station, made using the electro‑telegraph, with observers changing ends. It was, and is, believed to be reliable. For transport he had two hired motor lorries, and observed at 68 stations (apart from Fyzabad) between 17 December and 8 March. Seven nights were lost on account of cloud.

Geoid section. Hyderabad (Deccan) to Agra. J B Mathur resumed his latitude observations along meridian 78º, but after completing seven stations he fell ill with typhoid. He was replaced by B L Gulatee, the Department's Mathematical Adviser, who observed at another 27 stations bringing the section as far north as Agra, and crossing the main east‑west section at a point about 140 miles south of Agra. Northwards from Agra, the section can be extended to the Himalayan foothills using numerous old latitude stations. J B Mathur had by then recovered. He resumed charge of the detachment, and observed 11 stations in Bihar to give a short meridional section in longitude 84ºfrom the main east‑west section to the Nepal frontier.

The two main geoid section lines were thus completed. Important additions waiting to be done were (a) from Mandalay southwards to the Kra Isthmus, which was completed by D R Crone in 1937‑38; (b) two east‑west lines across the Peninsula, one in the latitude of Bombay and one at that of Madras, both of which were completed by R B Mathur in 1935‑36; and (c) a large number of lines which could advantageously be observed to complete circuits, some of which have since been observed.

In April 1935 I went on leave to England.

Conclusion

Comparison with modern methods.

(a) It may perhaps be instructive to compare the methods of 1930‑35 with those of the present day, but we must recall that in 1930 the object was to improve an existing triangulation system, such as those which still form the basis of surveys in most developed countries. I do not here refer to the methods which might form the basis of new surveys in (say) Africa.

(b) The principal problem which had been in the minds of Indian geodesists since about 1900 was to get the Indian triangulation computed on a world reference system instead of on our own arbitrary origin at Kalianpur. For this, two things were necessary. Firstly, we had to compute consistently in our own system, which among other things demanded the reduction of measured distances to spheroid level. One of the objects of our geoid sections was to provide data for this, and secondly, we had to locate the earth's centre of mass with reference to our own system. In 1930 this was the much more difficult problem, the first practicable step towards it being the interconnection of our system with that of other countries in order to form the largest possible block of homogeneous survey. Now, of course, the second aim is soluble by Doppler or other satellite fixes, and satellites can probably even be used to provide internal control in fairly large systems.

(c) In 1930 India had only 10 baselines, to which we added seven. Not really enough, but something to go on with. Now, the accuracy of geodimeter or other electro‑optical instruments equals that of invar baselines, and the increased range eliminates the old base extension net, which was the chief source of error. The cost is also immensely less. An old triangulation net is nowadays strengthened by the direct measurement of a substantial number of its sides.

(d) Azimuth control presents a similar problem. In 1930 India had 20 Laplace stations, but astronomical azimuth (without longitude) had been observed at more than 250 stations. In 1930‑35 we added more longitude observations at old azimuth stations whenever opportunity arose. Our geoid sections provided many such opportunities, and by 1938 we had a total of 44 Laplace stations, of which 11 were to some extent below the usual standards, but were good enough to be useful. It may quite soon be possible to get useful azimuth control from pairs of satellite fixes about 100 km apart.

(e) New triangulation chains to close circuits, or for international connections, would now be replaced by primary traverses, using either optical or microwave electronic instruments. In the late 1930s we were planning a 200 mile invar traverse (with baseline accuracy slightly reduced) to go eastwards from Calcutta, but the war upset our plans.

(f) For the observation of astronomical latitudes and longitudes, for which we mostly used prismatic astrolabes, the currently most promising instruments seem to be portable zenith cameras, which work fast, eliminate personal equation, and have random errors sufficiently small for measuring the deviation of the vertical.

(g) Bullock carts, camels, mules and coolies, can in many places now be superseded by motor transport or helicopters, but there must still be a number of situations in which either topography or finance impose the adoption of more primitive methods.

References. Full details of each year's geodetic work in India have been published in the Department's annual Geodetic Reports, 1922‑52. Some statistics and other details are in Professional Paper 28, The readjustment of the Indian triangulation, 1939.