THE 1942 W AIRARAPA, NEW ZEALAND, EARTHQUAKES: ANALYSIS OF OBSERVATIONAL AND INSTRUMENTAL DATA

In 1942, two large earthquakes, on June 24 (Mw7.2) and August 1 (UT) (Mw6.8), strongly shook the lower North Island, causing widespread moderate to severe damage. A third earthquake (M,6.0) occurred in the same area on December 2. These earthquakes have now been studied in detail by re-analysing seismograms from 1942 and by the collection and analysis of contemporary technical information and descriptive accounts from many sources. Results include new locations for the three main earthquakes and other moderate magnitude earthquakes in the sequence, summaries of building, lifelines and ground damage, new isoseismal maps and maps showing the distribution of landslides, liquefaction and other ground damage. The study has provided valuable information on the performance of buildings and lifelines in urban and small town environments at high intensities (MM8) and on the distribution of damaged buildings in central Wellington in relation to published ground shaking hazard microzoning maps and foreshore reclamation units. An important result is that scarp-like features described after the June earthquake as surface fault rupture are probably landslide-related rather than tectonically produced. This result and the lack of evidence for any other surface fault rupture, the closeness in time and space of the earthquakes both within the sequence and with the 1934 Pahiatua earthquake, and the similarity of the sequence to the 1990 Weber earthquakes have important implications for seismic hazard assessment of this part of the Hikurangi Margin. 125

A national network of eleven seismograph stations (plus a private station at Dunedin) (Figure I a) meant that the main events in 1942 could be located at the time using graphical means.All were found to originate from the same location near Masterton in the Wairarapa district [3,4] (Figure I a).The June and December mainshocks were found to be "shallow", while the August mainshock was recognised as a deeper event [3J.The extent of building damage in Masterton and of ground damage near Tauweru to the east of Masterton (Figure 1 b ), and the discovery of what was interpreted to be a surface fault rupture a few kilometres north of Tauweru on June 29, confirmed the location of the June event for the seismologists and geologists of the day [3-7J.earthquakes caused a scarp to form on what has subsequently been identified as a small fault in the Aorangi Ranges [9], over 60 km from Masterton and Tauweru.Dendrochronological dating confirmed that the growth of trees near the fault was disturbed within a year or two of 1942 [9].The occurrence of the 1942 earthquakes during the Second World War years when labour and resources were directed elsewhere meant that the Seismological Observatory published no comprehensive account of the effects of the earthquakes nor their seismological aspects in. the scientific literature, as had been done soon after the large earthquakes from June 1929 -1934 [for example , 10, 11].Hayes wrote only brief accounts of the building and ground damage [3,5], based on newspaper accounts and on observations during a 5day reconnaissance trip to the Wairarapa immediately after the June earthquake with Dr Lillie and M. Ongley (New Zealand Geological Survey).Ongley [6] elaborated on the geological aspects of the June earthquake, including his field observations of ground damage and suspected "fault rupture".Only one account of the damage was published by a contemporary engineering observer.Aked's [ 12] analysis of damage to buildings in the City of Wellington was published in 1945.An unpublished report [ I 3] by Luke, a Wellington City Council Engineer, covers much the same material as Aked, but the report also includes broad details on the distribution of chimney damage.In 1960, Johnston [ 14 J made some useful comments in a brief discussion of buildmgs damaged, again in Wellington only, from his investigation of 900 buildings in the central city area.In conjunction with other geological, historical and modern instrumental data, these reports were used to help delineate areas in Wellington that might experience shaking enhancement or microzoning effects in future earthquakes [15][16][17].
lsoseismal maps for the earthquakes were published in 1943 [4], but were very small scale, did not show individual intensity observations and were based on the Rossi-Fore! (R-F) scale.Various versions of the MM scale have been in use in New Zealand since January 1943 [GNS files].The most recent version is that of Dowrick (18].Eiby prepared new isoseismal maps for the 1942 June and August mainshocks based on R-F to MM scale conversion [19], so they could be used in attenuation of intensity studies.Although used for early attenuation studies [20,21], these maps are now being replaced with new MM intensity maps based on comprehensive studies [for example, 22-24, this study].These have been used to develop the most recent attenuation equations [25]. Studies based on instrumental records include statistical analysis of the aftershocks [2], and source mechanisms and depths of the June and August mainshocks from analysis of teleseismic wave forms recorded at Pasadena, California [26).Teleseismic body wave modelling studies of large historical North Island, New Zealand, earthquakes by Webb & Doser [27), are in progress.Their studies will yield source mechanisms, moment magnitudes and depths for the June mainshock and the two August 1 earthquakes [pers.comm.].
Previous papers by the authors and others [22][23][24]28] have demonstrated the benefits of comprehensive studies of large historical earthquakes for which there is short period local instrumental data and/or teleseismic data to establish locations, source mechanisms, depths, and surface wave and moment magnitudes.These investigations were undertaken to provide essential data for seismic hazard and intensity attenuation studies, data that had not been provided by more recent earthquakes because few large earthquakes have occurred from 1943-present, and few of the earthquakes that have occurred have been close to large towns.The purpose of the studies is to provide well-constrained isoseismal maps with well-determined epicentral parameters and source mechanisms, and to provide extensive data on the performance of structures and services in high density residential and business environments at moderate to high intensities and near source.Not only did the 1942 earthquakes occur very close to the town of Masterton, where there were many substantial non-domestic brick and concrete buildings, but also they provide the opportunity to document shaking-enhancement effects of a large distant shallow earthquake on reclaimed inner-city areas such as exist in Wellington.Further, the 1942 earthquakes occurred on the east coast of the North Island, an area which experiences moderate to high seismicity as a result of subduction of the Pacific Plate beneath the Australian Plate along the Hikurangi Trough (Figure la).Identifying the mechanisms and locations of these earthquakes within this zone is important for understanding how the relative movement of the plates is accommodated.
In this paper, the authors provide new instrumentally-derived locations for the 1942 Wairarapa earthquakes, their aftershocks and the two shocks that preceded the main events in June and August, new descriptive accounts and isoseismal maps, a detailed summary of lifelines damage, building damage at intensities of MMS, and a new map and description of the distribution of ground damage.Because of its importance in microzoning studies, Wellington City damage is summarised, and new information on the distribution of damaged commercial buildings is discussed.The lack of an identified surface fault rupture, the closeness in space and time of the earthquakes, both within the sequence and with the 1934 M,7.6 Pahiatua earthquake, and the similarity of the sequence to the 1990 Weber earthquakes, are among the results discussed that have important implications for seismic hazard assessment of this part of New Zealand.

INSTRUMENTAL DATA
According to Eiby [29], the New Zealand Network of eleven stations (plus a private station at Dunedin) (Figure I a) and the knowledge of New Zealand's crustal structure and velocities were developed enough by the early-mid l 940s that epicentres of magnitude 4.0 and above earthquakes between latitudes 38°S and 42°S could be calculated.An important proviso was that all stations were operating.

127
Howeve( rtie understanding of New Zealand's crustal and velocity structures were far less then than today, and moreover, epicentres could only be determined graphically. Based on the successful use of old records to relocate the 1934 M,7.6 Pahiatua earthquake and other March 1934 earthquakes [24], the authors have re-read phase arrivals from as many seismograms of the larger 1942 earthquakes as could be located and analysed them using modem location methods and velocity models (for description, see [30]).
Although instrumentation had improved over that of 1934 with the installation of a Wood-Anderson at Tuai (TUA) and an Imamura strong motion recorder at Wellington, at least six stations still had no absolute timing, or had poor timing, and recording speeds that were too fast (commented on in (24]).
Having New Plymouth (NPZ) Wood-Anderson records, Wellington (WEL) Imamura strong motion records and, in December, Imamura records from Bunnythorpe (BUN), none of which were available in March 1934, should have meant that the l 942 solut1011s would be better constrained than those for the 1934 earthquakes.However, while the network was adequate for locating isolated shocks the authors found that the resolution on individual records was not good enough for resolving locations when multiple shocks and/or very large shocks occurred.
Phase arrivals for the mainshocks, the shocks preceding the mainshocks and significant aftershocks were read from North Island seismograrns that could be found in the Seismological Observatory's archives (i.e.all except TUA).Reliable amplitudes, and S-P intervals (time between the arrival of the P-and S-waves) oc P and S arrivals, at other stations were added from readings tabulated by Eiby from the analysts' working books (Eiby, working papers in GNS files).Reread arrival times differ from l 942 readings by up to 3sec., some of the differences being attributable to the better magnifying equipment, which allows one more significant figure to be read than done in 1942.
Locations for the mainshocks, two shocks preceding the June and August mainshocks and larger aftershocks (ML.;,: 4.9) calculated from these data are shown in Table I.The estimated location error for any of the earthquakes is about 20 km.Because of the few stations constraining the epicentres, most are sensitive to the omission of any of the key stations (WEL, NPZ, HAS, BUN).Mw (Doser & Webb, pers.comm., June 1999) and M, [l] are given where available, while local magnitudes (Md of other located earthquakes are derived from the WEL, CHR (using 1942 readings) and NPZ Wood-Anderson records.
Epicentres of the June, Al.lgu.'it and December main.shocks and t.hei.rpreceding ,hocks: Except for the first earthquake on August l, all the locatable earthquakes in the sequence, including major aftershocks, are found to cluster about Masterton and within about 35 km of each other (Figure 2).They are also west of, and within about I 9--42 km of, the epicentres calculated in 1 942 [ see 3,4].The earthquake at 8.14pm on the evening of June 24 was a shallow, moderate magnitude ML5.3 earthquake, too small to be recorded teleseismically.The local station solution fits phase arrivals, including later crustal phases (for example, Pg), fairly well.For CHR and TAK the first arrivals were taken to be P* rather than Pn.Because of its unreliability (no absolute timing and poor quality record), BUN is not included in the solution, although including it changes the epicentre by about 5 km only, from close to the December epicentre to nearer the June and August mainshock epicentres.The epicentre would still lie within the cluster, however.Hence, because of its close relationship in time and space, the earthquake can truly be considered as a foreshock of the earthquake that occurred 3 hours later, and it will be referred to as such for the rest of this paper.Its location is compatible with the highest intensity of MM5 occurring in Masterton (see Table 5).
Using the WEL Imamura records, the June 24 mainshock can be separated into two major sub-events six seconds apart, the first sub-event being about magnitude 6.4-6.5, based on the ratio of its S-wave amplitude with that of the following subevent, which is presumed to be Mw7.2.Doser & Webb's body wave modelling (pers.comm., June 1999) also indicates two sub-events separated by about six seconds.The authors have attempted to locate both events using the local records, and the depth given by Doser & Webb (pers.comm.), but separating the events on stations other than WEL is difficult.
The solutions for the events are highly reliant on the depth chosen and on which stations are included, and vary by up to 15 km.The solutions listed are considered to fit the data best, including crustal phases, and the first sub-event is more reliably located than the second sub-event.In the following sections the main shock is referred to as though it were a single event at the location of the first sub-event ( which is within 5 km of the second sub-event).
The Mw5.6 earthquake on the afternoon of August I, previously thought to be in the same area as the June and August mainshocks, seems to be two earthquakes a few seconds apart at different locations.Neither of the earthquakes can be located instrumentally.The first earthquake is probably related to the June sequence.The location of the second, and larger, of the two earthquakes is identified by felt intensity data, that is, by the fall of many chimneys at Makuri and Coonoor, the fall of a few chimneys at Pongaroa, and by an anomalous large amplitude on HAS.
The HAS record is similar lo that of a 1934 Pahiatua earthquake aftershock (24] that was located about 15 km south east of Makuri.Therefore, the second earthquake appears to have originated about 50 km north east of the June mainshock and the Mw6.8 earthquake that occurred some eight hours later, and hence is not a foreshock of the latter event.About an hour after the Mw5.6 earthquake near Makuri (now to be referred to as the Makuri earthquake), a small earthquake (ML ~3.The inclusion of TUA moves the epicentre well to the east, but as the timing on TUA appears to be incorrect by 3.5-4 sec.for the well-constrained December shock and is not available for the June shock, the TUA P arrival has not been used to derive the location. The December mainshock epicentre is reasonably wellconstrained.A reliable S-P interval from the BUN Imamura contributes stability to the solution whereas for the earlier earthquakes BUN had only Milne-Jaggar records, which are difficult to read accurately and overloaded within 5-8sec.A depth of 12 km is obtained from the free depth solution, but the small area over which MM7 was experienced suggests the depth was greater.Varying the depth from 12 to 30 km changes the epicentre by at most IO km, the epicentre remaining within the highest intensity isoseismal (see Figure 5).The best fit location is within five kilometres of the June foreshock, although probably at greater depth than this event, and within 20 km of the June and August mainshocks.
Teleseismic data for June, August and December mainshocks:

Aftershocks:
Six aftershocks with magnitudes ML~ 4.8 occurred in the month following the June earthquake.Four were able to be located and all were within 25 km of each other and within 25 km of the mainshock (Figure 2).Their S-P intervals on WEL varied from 9.2-10.2sec.whereas the mainshock S-P interval on WEL was 10.7 sec.and the foreshock, 10.4 sec.).The range spanned by the S-P intervals at Wellington of all recognisable aftershocks within the first 24 hours after the earthquake was 8.2-12.0sec.The extremes of S-P intervals, that is, 8.2 sec.and 12.0 sec, both occurred within 1.5 hours of the mainshock, and the range does not change significantly up to the time of the August foreshock [2], nor after it.
Although there are too few locatable aftershocks to define the strike of the rupture zone, the zone encompassing them and the June mainshock is compatible with the preliminary source mechanism of Doser & Webb referred to above.Because of the orientation of Doser & Webb's strike with respect to Wellington, the range of S-P intervals (] sec equates to about 8 km) on Wellington records probably does not fully reflect the length of the rupture.Hence the minimum length of the fault rupture zone is estimated to be no less than about 35 km and probably no more than about 45 km.
According to Gibowicz [2], 644 aftershocks (ML ~ 2.9) occurred up to July 31.Three aftershocks reached magnitude 5 or greater (June 24 1309 ML5.3, June 24 2031 ML5.0, Jun 29 0702 ML5.I).The August mainshock initiated a small aftershock sequence at a rate of about 7 per day (ML~ 2.9) on the first day, compared to a rate of 82 per day on the first day after the June mainshock and 28 per day after the December mainshock [2].As no aftershocks of the August mainshock exceeded magnitude 5.0, and hence none were locatable, it is not known whether the aftershocks were at the same depth as the mainshock, or whether the aftershocks represent a reactivation of the June rupture zone.
The December mainshock initiated its own sequence of aftershocks (about 250 in 58 days with ML~ 2.9).None of the three largest events (all ML4.7) could be located.The WEL S-P intervals for the aftershocks to the end of December spanned 8-11 sec., most of them being between 9-10 sec.[2].The June foreshock also initiated a small aftershock sequence with ten events (3.3 ~ ML ~ 2.9) with similar S-P intervals (I 0.4 sec.± 0.5 sec.)occurring in the three hours between the foreshock and mainshock.

DAMAGE AND INTENSITY DISTRIBUTION
Collecting felt reports from Post Office and lighthouse operators, and newspaper clippings on an earthquake's effects was still an important task of the Seismological Observatory in 1942.Improvements in the instrumental network meant that intensity assessments were less necessary for locating epicentres of large earthquakes than previously, but information on building damage was more needed for the development of better building standards, as it is today.Unfortunately the early intensity questionnaires or "felt reports" were not well formulated and generally only those that included comments are now useful.
For this study, the newspaper collection was augmented and made as complete as possible.Relevant city and district council archives, National Archives, Government Department files, School Jubilee Booklets and records and reports of the Wellington Port Authority were searched for information and photographs.Some oral history tapes at Wairarapa Archive were listened to (enough to realise that gathering information from this source was time consuming and not very useful!) and eyewitness accounts sought from several people, and from diary accounts in the Alexander Turnbull Library.Public Works Department Engineers' building-by-building reports for all the larger Wairarapa towns significantly damaged in the earthquake were found at the Wairarapa Archive.The detail on buildings will be analysed in greater depth in a subsequent paper (by Dowrick).Another resource is a set of 420 building-bybuilding reports of damage to Wellington buildings (at Wellington City Council Archives).
Using these resources, new isoseismal maps have been compiled for the 1942 June 24, August l and December 2 mainshocks (Figures 3-5).Intensities assigned for these shocks, the June foreshock and the Makuri earthquake are summarised in Tables 2-6.Intensities were assigned using the 1996 modifications of the Modified Mercalli scale for New Zealand proposed by Dowrick [18].

THE Mw 7.2 JUNE 1942 EARTHQUAKE
The earthquake of June 24 was shallow enough (centroid depth hc=l2 km) and, with a magnitude of Mw7.2, laree enough to cause considerable damage, despite being centr;d away from large cities.It was felt over a large part of the North and South Islands, from Auckland to Dunedin (Figure 3, Table 2).The zone of heaviest damage, inside the MM8 isoseismal, covered an area of about 4,000 km 2 and encompassed six small towns and a total population in 1942 of about 24,000 people.The intensity reached a maximum of MM9 at a few locations near Masterton.
Inside the MM8 zone there were only two casualties, despite the fact that parts of about 50 brick buildings and thousands of brick chimneys fell and many windows were smashed.In Masterton, two were injured, one with a cut arm, and the other was serious enough to be admitted to hospital.Outside the MM8 zone, one woman broke her leg when she jumped from a hotel window in Otaki, and one man died in Wellington (MM6 and MM7) from carbon monoxide poisoning as a result of damage to gas fittings (Coroner's report).Otherwise, only a few minor cuts or bruises were reported.The few deaths and injuries may be attributed to the earthquake occurring at 11. 16 pm when most people were inside timber houses.,,-5,6 " ' 7 7 7?
Somewhat surprisingly, only one serious earthquake-induced fire was reported, despite it being mid-winter and heating by open fires in brick fireplaces being common practice.Again, this may have occurred because of the lateness of the hour.
One farm homestead at Waihakeke near Carterton was destroyed by a fire thought to have resulted from earthquake damage to electrical wiring.In Wellington there were two chimney fires, but these and several other minor fires attributed to earthquake damage were extinguished before a major fire developed.
At a distance of over 80 km from the source of both the June 24 and August I earthquakes, Wellington was at the threshold of structural damage, experiencing local intensities of MM6 and MM7 in both events (Figures 3, 4, Tables 2, 3).Using the attenuation expression for peak ground accelerations of Zhao et al. [32], an expected mean PGA-0.08g is estimated for soil sites in Wellington for both events.Because of the low level of damage of any individual building in Wellington further discussion of damage in Wellington is left to the section on microzoning.Nevertheless, the total loss to the Wellington community was substantial because of the large number of brick buildings and brick chimneys involved and because of the cumulative Takaka -MM5 Waverley.MM6?
The Brothers lighthouse -MM5 Wellington effect of two earthquakes within a few weeks.Luke [ 13] estimated the total damage costs for buildings in Wellington City as: . .

Domestic buildings
Masterton Within the MM8 isoseismal (Figure 3) in 1942 there were about 6,000 houses, predominantly of timber-framed construction.While most of these houses sustained earthquake damage, most of the damage occurred to.or was caused by the fall of, unreinforced brick chimneys.
According to a report to the Masterton Borough Council made four weeks after the main shock, 13,125 chimneys were "destroyed" in the Wairarapa district.As most houses had two chimneys, it is reasonable to infer that most or all chimneys throughout this region fell, in agreement with press statements.
A few scattered instances of houses within the MM8 isoseismal being structurally damaged by racking, twisting or falling off their piles, suggest local intensities of MM9.

Non-domestic buildings
Most of the non-domestic buildings within the MM8 isoseismal were in the townships of Masterton, Carterton, Greytown, Martinborough, Eketahuna and Featherston.The buildings were of one, two or three storeys and were made of timber, brick or reinforced concrete, with some hybrids.Of the 64 or so original reinforced concrete buildings in Masterton, some had brick infill, and only two (State Theatre and Woolworth's shop) had been designed to resist earthquakes.Eight other buildings that had originally been of purely brick construction had been substantially retrofitted after the 1934 M,7.6 Pahiatua earthquake [24), for example.the Masterton Opera House illustrated in Figure 7.Most concrete buildings were undamaged.Three suffered moderate cracking, and a further ten or so were slightly cracked.Ironically, Woolworth's shop was badly damaged when its roof was destroyed by falling brickwork from a brick building next door.One of the three-storey and one of the two-storey buildings are shown in Figures 8 and 9 respectively.
In Carterton, there were at least 27 unreinforced brick nondomestic buildings, of which at least five had walls or gables collapse, and at least ten others were in a dangerously cracked condition.Of the 14 buildings known to have been of reinforced concrete, most were apparently undamaged, two were slightly cracked and one poorly reinforced one had moderate cracks.
Greytown had at least six non-domestic brick buildings, in three of which the front facades collapsed.The others suffered lesser damage, especially the brick hospital building, which suffered mainly from chimney damage.The four reinforced concrete buildings, notably the 1937 Buchanan Ward at the hospital, appear to have been essentially undamaged.
Eketahuna had at least 21 brick buildings.Four of these buildings were so badly damaged that they were subsequently pulled down, while the remainder required moderate to substantial repairs.Damage to Eketahuna' s only concrete building, the County Council Chambers, was restricted to slight cracking of its parapet.
Martinborough had at least 13 brick buildings, all of which 137 required moderate to substantial repairs for cracked walls and parapets.Of the eleven reinforced concrete buildings nine were essentially undamaged and two suffered some cracks, but damage was only moderate.Dowrick, 1998).
•  Featherston had four brick non-domestic buildings which had damage ranging from a cracked wall to the fall of the front gable at Wilton's Garage.The eight reinforced concrete buildings suffered little or no damage.
It is evident that very few buildings in the zone of strongest shaking had been designed for earthquake resistance, despite the fact that about 30% of the concrete buildings post-dated the introduction of New Zealand's first earthquake loadings standard in December 1935, and despite the availability of the first draft of that standard from June 1931 [35].Indeed, at the time of the earthquake in 1942, of the above six towns, only Greytown had adopted the loadings standard as a bylaw, despite urgings to do so by (at least) the Masterton Borough Engineer, CR Mabson [36).
The campaigning for greater earthquake resistance, notably by Mabson in the Wairarapa, is evident in the retrofitting of several brick buildings in Masterton after the 1934 Pahiatua earthquake, and documented in his office's attempts to deal with earthquake risk buildings for at least 20 years after the 1942 earthquakes.This campaign involved leading engineers in Wellington, but it is sobering to note that the campaign did not bear any fruit at the legislative level until 1979 [37).
The good performance of all of the 90+ early brittle reinforced concrete buildings in the Wairarapa in the moderate to strong ground shaking of the main shock, despite almost all of them not being designed for earthquake, is clearly significant, but not exceptional.It is consistent with the similarly good performance of the pre-code concrete buildings in the M,7.8 1931 Hawke's Bay earthquake, in which such buildings also performed very well, even the 70+ concrete buildings in the near source towns of Napier and Hastings [23,38].In these towns the mean PGA's are expected to have been 0.5-0.Sg depending on ground class, and the intensity was MMIO [23] as indicated by the heavy damage to brick buildings.The success of all of these concrete buildings appears to be related to the fact that they generally had either concrete walls or brick infill walls in two orthogonal directions even though frequently not symmetrically arranged in plan (Figures 8 and 9).

Damage to Lifelines in the June 241942 Earthquake:
Electricity Supply Loss of electricity was experienced at many locations in the lower North Island from Norsewood (MM6) and Wanganui (MM6) to Wellington (MM6 and MM7).Power was not restored to parts of Greytown (MM8) for 30 hours, but elsewhere it was mostly restored within minutes to a few hours.Nevertheless, moderate costs were incurred by the Wairarapa, Tararua and Horowhenua Electric Power Boards (EPBs).The Wairarapa EPB (within MM8 isoseismal) costs were approximately as follows: reticulation £3,100, inspections £2,500, Kourarau hydro-electric headworks (near Gladstone) £500, and buildings £650 (totalling about $400,000 in 1998 values).Reticulation repairs in the Wairarapa were completed by July 3, nine days after the main shock.Full inspections of all installations were expected to take some months.
The Tararua EPB experienced some broken reticulation lines, but damage to the main distribution was not extensive.Its biggest loss was caused by heavy damage to its two storey brick head-office building in Eketahuna (MM8), which had to be demolished.The Horowhenua Power Board also reported broken reticulation lines, both low tension and several high tension wires.At the Manawatu Heads (MM7, but MM8 probably reached as a microzone effect), many poles in the low lying areas tilted, with consequent damage to wires.One or two poles broke.An explosion and small fire occurred in the switch gear at Pukepapa Road power house near Marton (MM6), and twenty minor service lines were damaged in Wanganui (MM6).

Gas Supply
The town gas supply in Masterton (MM8) was disrupted by the destruction of the brick chimney stack at the Gas Works, but the gas mains were undamaged.A steel chimney was erected within three days, and normal mains pressure was restored on the June 28, less than four days after the earthquake.The main damage was to the Gas Works buildings, of which the brick fronts of the engine room and coal sheds were badly cracked.Carterton (MM8) was the only other town in the Wairarapa to have town gas, and its Gas Works and reticulation were undamaged in the June 24 earthquake.The brick chimney stack was cracked in the August I event (suggesting it may have been weakened on June 24), but gas supply was back to normal within 24 hours.
At Petone (MM7), a 9 inch (225 mm) gas main was cracked due to subsidence of 50-150 mm on the west approach to a pipe bridge across the Hutt River.

Water Supply and Drai11age
Within the intensity MM8 zone, where the urban areas are all on stiff alluvium, water supply and drainage systems, including private drains, were almost undamaged.The only substantial exception was the collapse of the 40,000 gallon (180 m') Masterton water tower, made of bricks, concrete and timber.Water was supplied to this header tank by a 6 inch (150 mm) water main, which was severed when the tank fell.Council staff managed to turn off the control valve on this pipe within a short time and water supply was partially restored the day after the main shock, presumably at reduced pressure.
Fourteen leaks were found in the Waimeha water supply mains at Waikanae Beach (MM7), but the reservoir was not damaged.At Palmerston North in the northern fringe of the MM7 zone, the water supply was interrupted by the fracture of water pipes, as was well documented by the City Engineer [39], as follows: "One of the filters at the reservoir at Tiritea [7  After some earthquakes, damage to underground pipes only becomes apparent months or years later.For example, damage attributed to the 1987 Edgecumbe earthquake was still being discovered in the drains in Edgecumbe in 1999.
In Lower Hutt several burst water mains were reported but just how much damage to underground services in 1942 became apparent later is not known.Aked (12] makes the following valuable remarks about Wellington: "Little damage was caused ro the water mains.The pipes are of cast iron jointed with lead.
"Damage ro 1he sewer and stormwater drains cannot be easily detected unless the damage is of a major character.No damage of this characler occurred.Minor damage due to small fractures or subsidence does not usually become evident immediately.
..Where minor ji-actures occur the liquid waste percolates away leaving the solids behind in the pipe with a resultant blockage.The blockage can be cleared with rods in the usual manner as for normal operational blockages.Repeated blockages at any one position may ultimately reveal a fractured or collapsed pipe, but it is impossible to determine whether the original fracture was due to earthquake or not.A noticeable increase in the number of such blockages during the last eighteen months is suggestive of damage due to the earthquakes a.few months previously." Damage to hot water services throughout the MM6-7 area meant that many homes were without hot water for some weeks.As many rural areas of the Wairarapa depended on tank water, broken joints on tanks and damage to tank stands left many homes without water.

Telephone and Telegraph
Communications were disrupted for some days due to damage to Post (& Telegraph) buildings in the Wairarapa district, and fallen or damaged overhead telephone lines in both the Wairarapa and Manawatu areas.In Masterton, telephone and power lines were brought down by falling facades of brick buildings.Two days after the main shock a number of telephones, mostly country ones, were still out of order, due to line faults.In Palmerston North some 400-600 subscribers reported faults.The total cost of damage to Post Office buildings was estimated at over £100,000 ($6 million, 1998 terms).The brick Post Office building in Eketahuna was so badly damaged that it was condemned, forcing the removal of the telephone exchange equipment to another building.

Railways
The railway line most strongly affected ran from Featherston via Masterton to Woodville, with about 80 km of the track lying within the MM8 isoseismal, including 30-40 km probably directly above the rupture zone (Figure 3).Peak ground accelerations probably ranged from 0.3-0.6g.Following the clearance of a small slip from the line in the Rimutaka Ranges south of Featherston, the train service between Masterton and Wellington operated normally the day after the main shock.North of Masterton, two bridges were damaged and a section of track was out of alignment.This damage was apparently not heavy, but sufficient to keep the line between Masterton and Eketahuna closed for several days.
There was also some damage to the Wellington-Palmerston North line from a slip at Plimmerton (cleared by the next morning) and from subsidence of 29 lengths of rail between Paraparaumu and Te Hora.Temporary repairs and clearance of the slip enabled trains to proceed with care within 12 hours of the earthquake.A large rock also fell in the Manawatu Gorge breaking one rail.

Roads
For several days after the main shock the newspapers reported many instances of damage to roads within the MM8 isoseismal (Figures 3, 10).This damage took the form of subsidence of fills, cracking and blockage or undermining by landslides.Landsliding was widespread in the steep hillcountry that dominates much of the MM8 zone (see following section).The larger landslides were documented by Ongley [6], but many more small slides occurred that were sufficient to block some country roads for many days.Subsidences of about a metre occurred on approach fills to two bridges, one on the Tauweru River (Masterton-Stronvar Road) and one on the Ruamahanga River near Kahutara.Just north of Greytown on the main road to Masterton, the approach to the bridge over the Waiohine River subsided about half a metre forcing a temporary road closure.On a side road nearby, the timber structure of the Matarawa bridge was reported to have developed waves, suggesting differential settlement of its piers.About 15 km east of Greytown ground damage was reported to be heavy with cracks in roads.According to a report in The Dominion newspaper (20 August 1942) the Main Highways Board estimated costs of road repairs in the Wairarapa to be £20-25,000 ($1.2-l.5 million in 1998 values).
In the Manawatu, road damage was mainly confined to subsidence of bridge approaches and the cracking and subsidence of several roads, mainly to the west of Shannon and near Foxton Beach, where there were also extensive liquefaction effects (see following section).Several moderate to large slides occurred on the Shannon-Mangahao and Otaki Gorge roads.
As mentioned in the section on gas reticulation, there was slight subsidence of the approaches to two bridges in the lower Hutt Valley area, the pipe bridge over the Hutt River (50-150 mm) and the Seaview Road bridge over the Waiwhetu Stream ("very slight").In Wellington, minor cracking of the bitumen at the Chaffers Street/Herd Road intersection and at the approach to the Kelburn viaduct caused no problems for traffic.

Landslides, Ground Damage and Liquefaction Effects, June 24 1942:
The June 24 earthquake caused significant, widespread landsliding and liquefaction-induced ground damage throughout an area of about 11,500 km 2 in the lower North Island centred near Masterton (Figure l 0).This map shows the approximate locations and distribution of landslides, rock falls, liquefaction effects (sand and water ejections, lateral spreading), cracking and subsidence of roads, railway lines, and bridge approaches that were reported in newspapers, other archival sources or in [6].
The total area in which landsliding was reported was about 6,500 km 2 .Strong shaking caused slope and road cut failures, embankment cracking, and subsidence at considerable distance from the epicentre.The most distant slope failures reported were in the Pohangina and Apiti areas (about 85 to I 05 km north of the epicentre), and at Plimmerton and along the Hutt Road (about 70 km southwest of the epicentre).
The most extensive landsliding occurred in low-strength Tertiary rocks in a relatively narrow belt of Wairarapa hill country east of Masterton, extending about 60 km from Hinakura in the south to Bideford in the north, with areas about Tauweru, Gladstone to Longbush, and Martinborough to Hinakura particularly badly affected.Most of the landslides were of small to moderate size ( 10 3 -10-' 111 3 ), with many smaller rock and debris falls scattered throughout the hills.Only one or two large to very large landslides ( 10-5 -10 6 111 3 or greater) were reported (for example, at Spooners, ca.200,000 m\ Most roads lo the east and south of Masterton (to Tauweru, Gladstone, Longbush, Stronvar, Bideford) were blocked by landslides.
According to the County Overseer (Mr Wallace) the worst landslide damage in the Carterton area was at Gladstone where there were numerous landslides, as well as sand ejections, and fissures formed parallel to the Tauweru River [6].Between Gladstone and Longbush, slides were mainly shallow slips down dip-slopes in soft mudstone, falling into the Whangaehu River from both sides.
In the Masterton area, the landslide damage was worst near Tauweru.Here, hills were reported as being 'blown apart', with major slumping in mudstone and lateral spreading on the bank of the Tauweru River [6].A scarp feature formed on the side of a hill about 3 km north-northeast of Tauweru, on the southern side of the Mangatopitopi Stream, was identified by Ongley [6) as the surface fault trace associated with the June earthquake.However, recent examination by two of the authors has shown that this feature is actually the head scarp of a very large ( 10 6 m 3 or greater) incipient landslide (Tauweru Landslide, "l" in Figure I 0).This and another nearby incipient landslide feature are described more fully in the next section.Nearby, the lower Mangatopitopi Stream was dammed by a moderately large (ca.450,000 m 3 ) landslide (Mangatopitopi Landslide, "2" in Figure l 0).This small landslide-dammed lake was the only such feature reported following the earthquake, but there were probably others that could now be identified on 1943-44 aerial photos.
Many landslides were also reported from around the Eketahuna area and also along the Makuri Gorge road.Larger landslides affected the railway line between Mangamahoe and Mangatainoka north of Eketahuna.
In the Wellington region, some moderate to large slides occurred in greywacke in the southern Tararua Range southeast of Otaki, and very small to small debris slides and rockfalls occurred in the Wellington City and Kapiti coast areas.A landslide at Goat Point south of Plimmerton blocked both lines of the North Island Main Trunk railway line, and smaller rockfalls were reported on the line between Plimmerton and Paekakariki.On State Highway 2 (SH2), rock falls were reported along the Western Hutt road and there were small rock and soil falls from cut slopes on Mangaroa Hill.Rock falls, cracking and subsidence (fill failures near summit) occurred on the Rimutaka Hill road, but it remained negotiable with care.A small failure was reported on the Rimutaka railway incline (Cross Creek), and there were also slides in the Waiohine River gorge west of Masterton.In the upper Otaki River area Lake Orewa (a ridge-rent tam) was reported after the earthquake to have been drained, probably due to ridge-top cracking.Further to the north, boulders fell on to the road in the Manawatu Gorge.
Liquefaction effects were evident in many places, with sand and water ejections (sand boils), cracking and minor lateral spreading widely reported (Figure I 0).These were far more extensive than reported by Fairless and Berrill [40] in 1984.
In the Wairarapa, liquefaction-related cracking badly affected riverbanks and stopbanks, particularly around Gladstone and Greytown, and sand ejections were reported from as far away as Lake Ferry (70 km southwest of the epicentre).
In the Manawatu, the Foxton Beach area and Manawatu River stopbanks (50-65 km northwest of the mainshock epicentre) were badly affected with widespread cracking and sand ejections (Figure 11 ).At the Manawatu Heads and Foxton Beach practically all power poles in swampy, lowlying areas were tilted, and one fell over, due to liquefactioninduced ground failure.Although at 60 km or so from the earthquake source, the alluvial and flax swamp deposits present in these areas are highly susceptible to liquefaction effects, as demonstrated by other historical earthquakes such as the 1855 Wairarapa [41,42] and the 1934 Pahiatua [24,41] earthquakes.
Sand boils also occurred near Aotea Quay in Wellington (over 80 km southwest of the mainshock epicentre).

Relationships between ground damage an,d MM intensity
The isoseismals shown in Figures 3 and I 0, generally assigned on the basis of building and contents damage, are consistent with incidence of landsliding and ground damage.The heaviest ground damage, where landsliding and blockage of roads was common, occurred within the MM8 isoseismal.Within the MM7 isoseismal area, landslides were sparser, and rare on natural slopes, but many cut slopes and fills failed.In the MM6 isoseismal area, minor rockfalls or falling rocks occurred.These were mainly of little more significance than nuisance value, except at Pohangina and Apiti , about 90 and ll O km north of the epicentre, where two significant failure areas were reported.The moderately large landslide at Goat Point near Plimmerton, (MM7) occurred on a cut slope that had a prior history of instability.The same area failed again during rain in the late 1970s, requiring remedial works to protect State Highway 1.
Liquefaction in the Gladstone and Dalefield area is consistent with MM8 shaking.The extensive liquefaction effects in the more distant Shannon and Foxton areas (within the MM7 isoseismal) are also consistent with MM8 shaking, but this is probably a microzone effect caused by highly susceptible ground conditions in these areas.

Relationship of landsliding to epicentre and aftershocks
A recent study of historical earthquake-induced landsliding in New Zealand [42] showed a good correlation between the areas most affected by landsliding during earthquakes and the (subsurface) fault rupture zone, as indicated by the zone of aftershocks.Convincing close relationships were demonstrated for the 1929 Murchison, 1968Inangahua, 1990 Weber, and 1994 Arthur's Pass earthquakes.This association suggested that, for a given earthquake and allowing for topographic and terrain effects, landslide distribution could provide a reliable indication of the epicentre location, and also the extent of the fault rupture zone at depth.The same association is supported by this study of the June 24 1942 earthquake.The area of most intense landslide damage caused by the earthquake also shows a close correlation with the zone of aftershocks, and to the epicentre (Figure I 0).Topographic influence is also apparent, with most landsliding occurring in steeper hill country on the eastern side of the aftershock zone.Few landslides were reported on the western side of this zone, which extends across the Ruamahanga River plains, where ground damage was mainly limited to sparse ground cracking, embankment subsidences, and minor liquefaction effects.

Tectonic setting of the 1942 Earthquakes and swface rupture in the June 241942 earthquake:
Compared to much of the rest of the country, central New Zealand, which includes the location of the 1942 Wairarapa earthquakes, has a high rate of occurrence of large magnitude earthquakes, and consequently has a high seismic hazard (for example, [43][44][45]).The tectonic setting of central New Zealand, that is, the driving mechanism behind most of the larger earthquakes, is dominated by the oblique westward subduction of the Pacific Plate beneath the Australian Plate.In the southern North Island, the subduction margin between the two plates, and the overall structural of the deformed overriding Australian plate, has a general northeast-southwest trend.The relative motion of the Pacific Plate, with respect to the Australian Plate, is approximately due westwards (azimuth of ca 260°) at a rate of about 40 mm per year.
To a first approximation, the oblique motion between the two plates is partitioned into margin-parallel and margin-normal displacement in a manner first described by Fitch [46], and in the New Zealand setting by Walcott in 1978 [ 4 7].The margin-parallel motion appears to be largely accommodated by slip on a northeast-trending zone of active right-lateral strike-slip faults at the rear, northwestern boundary, of the subduction margin.At the latitude of the 1942 Wairarapa earthquakes, this zone of strike-slip faulting comprises, from east to west, the Wairarapa, Wellington, Ohariu, Pukerua and offshore Wairau faults (for example, [48]) (see Figure 2).The margin-normal motion, that is, contraction perpendicular to the margin, appears to be expressed as thrusting, reverse faulting, and folding east of the strike-slip faults, and as slip on the subduction interface (for example, [49][50][51][52], and references cited therein).There are, however, complications and exceptions to this simple model.For example, there is a group of active faults that extend east-northeast across the subduction margin from the Wairarapa fault near Carterton and Masterton (for example, Lensen [53]), the best expressed of these faults being the Carterton and Masterton faults (Figure 2).Recent work by Zachariasen et al [pers.comm.]suggests that the Carterton Fault is primarily right-lateral strike-slip with a lateral slip rate of at least 2 mm/yr.The rate of movement and sense of slip on the Masterton fault is still, however, not known.Also, there are no doubt elements of margin-normal contraction accommodated within the zone of strike-slip faulting, as well as margin-parallel motion accommodated in the "contractional" zone to the east.Two good historical examples of the latter are the 1934 Pahiatua earthquake, and the June 24 1942 Wairarapa earthquake.Both these earthquakes appear to be large, M >7, strike-slip earthquakes with a preferred focal plane that is steep-dipping, right-lateral, and north-northeast to northeast striking.

Surface fault rupture in the June 1942 earthquake
Immediately following the 1942 June 24 Wairarapa earthquake, a group of New Zealand Geological Survey geologists visited the epicentral region.Their observations regarding earthquake-triggered ground damage are reported in Ongley [6].Near the township of Tauweru, they report scarp-like ground damage features that they ascribe to primary surface fault rupture.These scarp-like features, with a combined length of about I km, are located west of the Tauweru River on both sides of Mangatopitopi Stream about 2.5 km north of the Carterton fault, and about 3.5 km northnortheast from Tauweru township (approximate grid references, NZMS 260 S27 465265 to 474268).Air photos (904/27, 28 and 29), taken only 14 months after the June 1942 earthquake, clearly show these scarp-like features, as well as quite a few pre-existing deep-seated landslides, some of which appear to have been freshly re-activated.The photos also show numerous shallow seated slides that were presumably the result of the heavy rains that followed the earthquake.
On June 8 1999, Gaye Downes and Russ Van Dissen visited and photographed (Figures 12, 13) the scarp-like ground damage features located southwest of Mangatopitopi Stream (Ongley's Figures 4-7; Ongley's Figures 4, 6 are reproduced here in Figures 12, 13).The authors could also look northeast across Mangatopitopi Stream to the scarp-like features depicted in Ongley's Figures 8 and 9. (Figure 9 of Ongley [6] is also reproduced as Figure I 69A of Cotton [54]).The scarplike features located southwest of Mangatopitopi Stream generally face south, have a strike of ca N80°E, and are on trend with and continuous with the head scarp of a large south-flowing pre-existing landslide.South of the scarp-like features, i. e. downslope of the features, the hill slope comprises classic hummocky landslide topography.
The scarp-like features located on the northeast side of Mangatopitopi Stream, as depicted in Figures 8 and 9 of Ongley [6] and seen in the background of Figure 13, have largely been destroyed as they now define the lateral margin of a narrow landslide.Originally these features had an overall northeast-trend, and faced northwest.It is interesting to note that the en-echelon pattern of surface cracking defined by the scarp-like features, shown in Figure 9 of Ongley [6], is consistent with the differential lateral movement expected across the lateral margin of a landslide.
Another important observation relevant to the genesis of the scarp-like ground damage features is that there is very little, if any, offset of relatively pronounced strike ridges of resistant lithology on either side of, and oriented approximately normal to, the western and eastern extent of the scarp-like features.That is, the scarp-like features are contained within the zone between these two parallel strike ridges, and these north-south trending strike-ridges are not displaced by the scarp-like features.In map view, the two strike ridges form the two vertical lines of the capital letter "H", and the scarp-like features form the horizontal line between.
It is our consideration that the scarp-like ground damage features reported by Ongley [6] as surface fault rupture associated with the June 1942 are in all likelihood landslide related.The scarp-like features are invariably associated with, and in most cases, coincident with, pre-existing or subsequent landslides, and there appears to be no offset of pronounced geological units along strike of, but at a high angle to, the scarp-like features.The authors consider that the best explanation for the formation of these scarp-like features, at least those located southwest of Mangatopitopi Stream, during the 1942 earthquake is not tectonic fault rupture of the ground surface, but rather minor re-activation of a preexisting landslide.The re-activation of the landslide was no doubt triggered by the strong ground shaking generated from the 1942 earthquake.Ongley [6] describes the maximum vertical displacement across the freshly formed scarp as approximately 3 feet, and notes that the freshly formed scarp comprises only a part of a larger scarp of about IO feet.Ongley proposed that the larger scarp height was the result of additional, older, fault movements.The authors agree with Ongley that the larger scarp height documents repeated, older, movement, but the authors ascribe the movement not to tectonic fault rupture, but rather to landslide failure.
Figure 12: Photographs of the scarp near Tauweru (Landslide 1 in Figure 10) that Ongley attributed to fault rupture, now recognised as the western side of the head scarp of a large landslide.The upper photograph was taken in June 1999 at approximately the same location as Ongley's [6] Figure 4 shown below.
Figure 13.Photographs of the scarp near Tauweru that Ongley attributed to surface fault rupture, no.w recognised as the eastern side of the head scarp of a large landslide.The upper photograph was taken in June 1999 at approximately the same location as Ongley's [6] Figure 6 shown below.
The authors are not the first to question the tectonic origin of the scarp-like features described by Ongley.For example, Lensen in 1968 [53] states: " ... those traces that were definite su,face breakages formed during the Wairarapa Earthquake of 24 June 1942 (Ongley, 1943) [6] Ongley ( 1943) instead represents the effects of ground shaking rather than fault slip." The lack of damage to water and sewerage systems in Masterton, through which the Masterton Fault passes, suggests that this fault did not rupture to the surface.The lack of a report of significant ground displacement across the Masterton-Tauweru Road, which would have been sighted by Ongley and others, and which crosses the Carterton Fault, suggests that this fault also did not rupture to the surface.In addition, the fact that intensities greater than MM9 occurred only at isolated locations, and were not widespread near the epicentre, is consistent with rupture failing to reach the surface.
No explanation can be found for reported faulting in the Aorangi Ranges (9) some 60-70 km south-south-west of the epicentres of the June and August earthquakes.A scarp was apparently noticed by a resident of the Pararaki Valley after a large earthquake in 1942, and reported in Berryman (9) and Stevens (56].Berryman [9] investigated the site, reporting that, although near the headscarp of a large pre-existing slip, the scarp suggested faulting rather than landsliding or ridge renting, as the observed throw was contrary to that of the headscarp of the landslide.Further, the trace extended over the spur of a ridge and is continuous for ca.400 m.Dendrochronological evidence suggested disturbance of growth rings in the trees that was consistent with its having occurred in 1942-43 [9].The site is too distant from the 1942 June, August or December mainshocks to be primary or even secondary faulting, and no other large earthquakes occurred nearby within the June-December 1942 period.However, it is within the MM7 isoseismal of the June earthquake and hence ground damage in susceptible terrain would not be unexpected.

THE Mw6,8 AUGUST 11942 EARTHQUAKE
While the August I earthquake (Mw6.8, he 40 km) was almost as large as the June 24 event (Mw 7 .2, he 12 km) and its epicentre was within a few kilometres, it was far less damaging overall because of its greater depth (Table 3 ).The strongest isoseismal is MM7 compared to MM8 in the June event (Figures 3, 4).Hence the damage to domestic and nondomestic buildings is insufficient to warrant describing it in as much detail as the June event.
A significant feature of the earthquake was the extensive additional damage caused to previously damaged structures and to freshly rebuilt (and hence, very weak) structures, particularly chimneys.This was more noticeable in areas only moderately damaged in the June earthquake, for example, in Wellington.Here, the collapse of buildings, and parapets in the central city area and the repeated destruction of many chimneys made the August earthquake seem much more severe than that in June.
In the Wairarapa, there was considerable further damage to unreinforced brick chimneys and some extra damage caused to some of the brick buildings in the townships, particularly Masterton, Carterton and Eketahuna.Some timber houses in rural areas were also reported as being damaged, e.g."badly wrenched" (the meaning of which is unclear).
There were no casualties reported.As with the June earthquake, the lack of casualties can be attributed to the time of the earthquake (00.34am local time), when few people were on the streets and most people were asleep in their timber houses.
The assignment of intensities of MM6 and greater are less certain than for the June earthquake because of the presence of previously damaged, or nearly repaired, structures.At locations where this effect is significant, intensities that might superficially be assigned as MM7 or MM8 are assigned as "MM7?" (Table 3; Figure 4).

Electricity
Loss of electrical supply affected fewer locations over a smaller area and for a shorter time than in the June earthquake.The longest period of power loss ( 18 hours) was in Eketahuna (MM7), while elsewhere power was generally restored in less than an hour.In Wellington, the electricity supply was interrupted for about 20 minutes, the failure being attributed to the tripping of a switch at Mangahao and the failure of an earth clamp on a transformer at Pauatahanui.The collapse of parapets and facades in Wellington (MM6 and MM7) also brought down power lines in several places.There was further damage to poles and service lines at Foxton Beach (MM7, possibly MM8 in places), where liquefaction effects were again evident.Several hundred service lines and several feeder lines required attention in the Manawatu and Horowhenua areas (MM7).

Gas Supply
Two instances of damage to gas supply were recorded.At the Carterton Gas Works (MM7), not damaged in June, there was minor damage to retort doors, the brick chimney stack and the engine room roof.The supply was back to normal within 24 hours.Some gas installations were damaged in Wellington's Manners Street.

Water Supply
Eketahuna (MM7) reported the most serious damage to water mains.The subsidence of a road in the town caused the water main to break in three places, and the town was without water for most of the day following the earthquake.Several connections to houses at Otaki (MM7) were damaged and one water main was broken near the reservoir servicing Palmerston North (MM7).In Wellington (MM6 and MM7), one or two water mains were damaged, and three joints in the water pipe from Wainuiomata sprang leaks.

Telephone and Telegraph
Telegraph services were not disrupted and the incidence of damage to telephone services was considerably less than in June.For example, in Palmerston North only 170 faults were reported, compared with over 600 in June.

Railways
Locations in the Wairarapa that caused problems for railway services in June were again troublesome, with landslides, and subsidence of bridge approaches and parts of the line.However, service along the section of the line between Eketahuna and the Rimutaka Incline (MM7), where the most serious delays were caused, was restored within 12-15 hours.Minor damage between Waikanae and Levin (MM7) meant that trains could travel only at reduced speed, as they could at Kakariki (MM6), near Halcombe, where the railway bridge was reported out of alignment.

Roads
Damage from landslides, subsidence of road edges• and bridge approaches (described in greater detail in the next section) was widespread in the Wairarapa, new slips and subsidences occurring at many of the locations that were badly affected in the June earthquake and also in several new locations within the highest intensity isoseismal, MM7.At Tinui and Mauriceville (both MM7), roads were cracked and bridges damaged whereas this had not been reported in June.At Eketahuna (MM7), subsidence of one road, unaffected in June, was serious enough to break the water main.Considerable damage was done to four bridges in the Wairarapa, with minor damage to a fifth, all within 10-15 km east to southeast of Masterton.The principal damage was to abutments, with no damage to the superstructure.The culvert at Devil's Elbow, near Tauweru, was shattered.Heavy rain and floods two weeks before the earthquake may have contributed to the landsliding and subsidence.
In the Wellington area (MM6 and MM7), there were superficial cracks in Upland Road and Glenmore Street, while in the Manawatu and Horowhenua area, there was further cracking of the roads about Foxton (MM7) and a small landslide and superficial cracks cast of Otaki (MM7).
Landslides, Ground Damage and Liquefaction Effects, August 1 st 1942: Earthquake-induced landsliding, ground cracking, subsidence and liquefaction effects caused by the August earthquake were less severe than those in the June earthquake and occurred over a significantly smaller area (5,600 km 2 ; Figure 14).Landslide damage occurred in a few localised, widely separated areas in the Wairarapa hillcountry to the south, east and north of the epicentre, but also to the west as far away as Pukerua Bay, Paekakariki, and Otaki.Many of the slope failures occurred in the same general areas as in the June earthquake, but the slides were generally far less numerous and damaging.
There was only moderate landslide damage in the Wairarapa, and minor subsidence of bridge approaches.The most intensive damage occurred in the Martinborough area, with all roads to the coast again blocked by slips, cracks, and subsidence.Slips again blocked the Stronvar road southeast of Masterton.In the north, slips fell on to the road south of Eketahuna, where there was also road cracking and subsidence.Again, there were numerous slips in Makuri Gorge.A slip occurred on the Rimutaka Incline between Cross Creek and the summit, and small landslides were reported in the Otaki Gorge area.The Paekakariki Hill road was closed by a "big slip" while at Pukerua Bay there was a "slight fall of earth" near the railway line.On the railway line south of Plimmerton there was no further trouble from the June slip area at Goat Point.There was little landsliding in Wellington City with only minor, very small regolith slides and falling rocks reported in a few places.
Soil liquefaction effects were also evident during this earthquake, particularly in the far field.Significant localised liquefaction effects were again reported in the Foxton to Levin area (Figure 14).Sand fountaining occurred along the river front at Foxton Beach, and bridge approaches in the area subsided.In the swampy ground in the Horowhenua area, power poles were tilted over, probably by liquefactioninduced ground failure.On the Paiaka Road to the Manawatu River (west of Shannon) widespread "sand geysers" again occurred over about 5 acres (2 ha), with sand piled up 400-500 mm and spread out over 2-4 m from some vents.Similar effects also occurred in about the same location during the June earthquake, although possibly more widespread.The widespread liquefaction effects in the Foxton-Horowhenua area indicate local shaking bordering on MM8 in an otherwise MM7 area, emphasising the high susceptibility of ground in the area to soil liquefaction and hence, microzone effects.Surface Fault Rupture in the 1942 August 1 Earthquake?

I I
Given the depth of the 1942 August earthquake (about 40 km) surface fault rupture would not be expected, and indeed, Hayes [3) in his annual report for the Dominion Observatory notes that "no fresh surface faulting was found on 2 nd August".However, Neef [8) suggests that a 2 km long bedding plane parallel fault scarp to the east of the Alfredton Fault, which he noted as fresh-looking on I 944 aerial photographs, ruptured in the August 1942 earthquake (the mistakenly identified fault trace for the June earthquake [6) having been located elsewhere).Based on their observations on the same 1944 aerial photographs, and on photograph~ taken in I 935 and 1947 of the main Alfredton fault, Schermer et al [55) observe that: "the main Alfredton fault trace is at least as fresh [as Neet"s bedding plane fault], and probably moved at the same time", and that-the most likely date of the most recent rupture along it was in 1855, not 1942 or I 934 (in the 1934 Pahiatua earthquake).

THE Ms6.0 DECEMBER 2 1942 EARTHQUAKE
The calculated epicentre of the December earthquake correlates reasonably well with the highest intensity area (Figure 5, Table 4), where each of the small communities yet again experienced the destruction of crockery and glassware, and the fall of at least a few chimneys.Some of the chimneys were still "green" from recent reconstruction after the previous earthquakes.The earthquake exacerbated previous damage elsewhere, particularly in Wellington, but none was serious.Some areas of the Wairarapa on alluvial flats (for example, at Tuturumuri) and in Wellington appear to have bee_n s_haken more strongly than others nearby, probably md1catmg a microzone effect.

FORESHOCK ON JUNE 24 1942 AND EARTHQUAKE ON AUGUST 1 1942 (THE MAKURI EARTHQUAKE)
With a magnitude of ML5.3, the earthquake on the evening of June 24 (3 hours before the Mw 7.2 earthquake) was insufficient to cause any damage other than the fall of a few items from shelves in Masterton (Table 5).
The Mw5.6 Makuri earthquake on August I was widely felt in the southern half of the North Island (Table 6).As this earthquake could not be located instrumentally an epicentre has been adopted that is consistent with the intensity data.The only significant damage was the fall of many chimneys at Makuri (MM7?) and Coonoor (MM7?) and the fall of a few at Pongaroa (MM6).As noted earlier, intensities have ~een assigned "1" where the effect of cumulative damage from the June earthquake is uncertain.

DAMAGE IN WELLINGTON IN THE JUNE 24 AND AUGUST 11942 MAINSHOCKS AND MICROZONING EFFECTS
The variability of damage from one location to another nearby has been recorded in the media in many New Zealand 151 earthquakes.In the 1942 earthquakes clear variations in ground shaking were evident in Wellington City and in the Manawatu.Wellington City has the best dataset for examining the microzoning effects in these events.Some damage information also exists for the Hutt Valley, but the distribution of houses damaged is not well determined.Other than newspaper accounts, the only major sources of information found were State Housing Corporation files (held at National Archives) detailing damage to State Houses in Lower Hutt.These indicate that a great many houses suffered chimney damage, some damage to plumbing and some minor cracking but the distribution of damage shows little more than the distribution of State housing in the Hutt Valley, i.e.Waterloo/Woburn, Avalon and Naenaeffaita.
In Wellington, depending on location, intensities of MM6 and MM7 were experienced in the June and in the August earthquake.The August earthquake appeared to do more damage, especially to commercial buildings.With only six weeks between the two large earthquakes repairs from the June event had not begun, were still in progress, or were recently completed with mortar or concrete still weak.Reports of the damage prepared by engineers of the time do not attempt to separate out the effects of June and August earthquakes, which they well recognised as cumulative.
The Wellington data can be divided into residential and co~mercial damage.The residential damage was primarily to chimneys, although several houses in Webb Street and Barker Street were so badly damaged from falling chimneys and cracked and fallen brick walls that they were evacuated.
Luke [ I 3) prepared a map of the percentage of damaged domestic chimneys in each of 32 building districts (reprinted in [15)).Chimney damage per district varied from Oto 84%.While there is a minor correlation with age of housing, probably due to improved building practices in the newer houses, there is a far better correlation with areas of soft sediment [ 15].
The overall pattern of damage is consistent with the 1992 ground shaking hazard map [ 16,57), except for Kelburn and Northland, which had 63% and 54% chimney damage respectively.Both suburbs are almost entirely in the lowest hazard zone on the 1992 map.As the original data used by Luke is not available, it is not possible to examine this discrepancy nor to recalculate the percentage damage in each of the 1992 zones.
Of the commercial buildings, 36 required major structural repairs and a few hundred others required minor repairs, according to the newspapers published a week or two after the June earthquake.While some parapets collapsed, many more were only cracked and the most obvious damage to the Central Business District (CBD) was the breakage of many hundreds of windows (over 300 in one building alone), the extensive stock damage, and the extensive cracking and tilting of brick walls.The Te Aro area was reported to be badly affected as were the areas about Cambridge Terrace, Webb and Buckle Streets.In the wharf area, there was subsidence of fill at several locations and liquefaction was reported near the Social Security building in Aotea Quay.
The damage to the CBD from the August earthquake was extensive and more obvious than in June, with many parapets, masonry walls and brick walls cracked and dangerous, or collapsed into the city streets.Although some instances of damage were reported to be in buildings not previously damaged, most were seen by the media and the building/engineering community to be the result of the weakening caused by the June earthquake.
Some new interpretation of the distribution of damage within the CBD is possible with new information on the damage to non-Government commercial structures.A map of unknown authorship (Figure 15a) was photocopied in about 1992 from the original held by the Wellington City Council (and now apparently lost).15a).Neither the construction type nor the extent of damage is specified.No text other than the map legend is available for the map so it has only been interpreted in broad terms and only in the areas where there are few residential structures, as damage to residential structures does not appear to be included on the map.
The authors have limited their analysis to the pre-1914 buildings, as the map clearly shows that these older buildings suffered the greatest percentage of damage.There is broad agreement between the distribution of buildings damaged and the 1992 earthquake shaking hazard zoning map previously referred to (Figure 15b), but there are two areas where the 1942 damage is higher than predicted by the zoning on the 1992 hazard zoning map.The concentration of damage around Manners Street falls in Zone 2 and a small concentration near the northern end of Lambton Quay is in Zone I.The sediments in the Manners Street area may be deeper than estimated in the 1992 sediment thickness map [58] as the subsurt•ace control there is limited (Perrin, pers comm).The buildings at the north end of Lambton Quay are on bedrock (the basis for the Zone I classification) and are the only buildings listed on the map in Zone I that were damaged.
Many of the buildings damaged in 1942 were sited on areas of fill seaward of the pre-1855 shoreline, the first such reclamation occurring in 1852 and the last major reclamation being completed in 1914.The boundaries of the 1992 shaking hazard map were based largely on the depth of soft sediment and did not differentiate between different units of fill.However, the newly found building damage map shows a strong correlation of damage to particular units of fill (Figure 15c).Damage incidence in individual units varies from zero to over 50% of the buildings.The difference may be due to a number of causes, including the type of fill and whether it had been compacted, perhaps the similarity of construction or use, or to the construction technique in vogue immediately following reclamation.Nevertheless, the effect merits further investigation.
The Wellington City Archives holds original building inspection records of over 420 buildings dating from the June 24 earthquake onwards, in which there are details of the type of building (though not the age), the damage to the structure and recommended repairs.A useful, separate study could be conducted to interpret the damage descriptions from these and the newspaper accounts, and to prepare a detailed microzoning report on the damage to commercial structures.Such a study could address the effect of building height, and type of structure as well as the microzoning effects.

IMMEDIATE RESPONSE AND RECOVERY
Without doubt, recovery and reconstruction after the 1942 earthquakes were impeded by the scarcity of labour and materials created by the Second World War.Repairs to chimneys after the June earthquake proceeded more slowly than many householders and local authorities wished.Heavy rain (and flooding) throughout the southern North Island on July 13-14 heightened awareness of inadequate progress.
Despite the restriction that one essential chimney only was repaired for each household and despite repairs being organised by local authorities rather than individuals, repairs proceeded slowly and many complained about overcharging and unnecessary demolition of chimneys by the Emergency Precautions Service (EPS) and the army.
On the other hand, the usefulness of having EPS personnel, which had been formed to manage and respond to war emergencies, was more than evident.An unknown writer in the Wairarapa Times-Age June 26 1942 writes: "The necessity of an EPS organised to function in peace time as well as war time has been brought home by the incidents of the last few days".The EPS personnel were mobilised and immediately available for managing public safety, demolishing dangerous buildings and chimneys and setting up canteens and shelters for labourers, and for small numbers of evacuees in Wellington and the Wairarapa, just as the regional and local authority Civil Defence units are today.
The military in the Wairarapa also provided labour for patrolling streets and for demolition work.Soldiers with building skills were relieved temporarily of their duties, primarily to help rebuild chimneys.
The 1942 earthquakes also provided the impetus to the Government to establish a national insurance scheme.The Earthquake and War Damage Commission was established in 1944.

DISCUSSION
The study has provided valuable information on the locations and effects of the 1942 earthquakes, all of which are relevant to seismic hazard assessment.However, several results have significant implications for seismic hazard assessment that takes into account the variability of hazard with time.
Although the June 1942 earthquake was shallow enough ( ~ 12 km) and large enough (Mw7,2) for surface rupture to have occurred, evidence suggests that it did not.Thus, there is a strong likelihood of other similar strike-slip earthquakes having occurred in the past without leaving a clear surface manifestation of their occurrence in the geological record.Hence, attempts to define the long-term rate of occurrence of similar or lesser magnitude earthquakes based on paleoseismic studies of the geological record are likely to underestimate the frequency with which large damaging earthquakes have occurred, leading to a potential underestimation of seismic hazard.
Also significant for seisrnic hazard assessment is the closeness i.

Figure 15 (c): Map of individual units of fill and dates of creation. Compiled using a 1936 Wellington Harbour Board map [59] and Baillie [60).
The closeness in time and space of the earthquakes within the June-December 1942 sequence also suggests some causal relationship and possible stress triggering.The occurrence of a sequence of moderate to large earthquakes like the June-December 1942 sequence is not unique in the Wairarapa-Hawke' s Bay area.In February 1990, a Mw6,2 earthquake in the subducted plate under Weber was followed by a Mw6.4 earthquake in the upper plate in May 1990 [64].Two subsequent events, both ML5.5, occurred in the subducted plate in August 1990 and March 1992.All were within 15 km of each other.The new epicentre and depth of the June 1942 mainshock show that it occurred in the upper plate while the source of the August 1942 earthquake was I 5 km to the southwest in the subducted plate, although it should be recognised that epicentres may be mislocated by up to 20 km.Nevertherless, the 1942 earthquakes seem to be paired in a similar manner to the 1990 Weber earthquakes, although in 1990 the first event was the deeper event and in the upper part of the subducted plate [64], somewhat shallower than the August 1942 event.This coupling of earthquakes (and possibly the occurrence of a third earthquake in 1942, and a third and fourth earthquake near Weber in 1990-1992) shows that such sequences are probably not isolated occurrences and the probability of such sequences should should be taken into account in hazard modelling.

CONCLUSIONS
This study has achieved its goal of providing more reliable and extensive data on the locations and the effects of the June 24 Mw7.2, August I Mw6.8 and December 2 M,6.0 1942 Wairarapa earthquakes, their major aftershocks, and other associated shocks.In particular, damage data provide valuable insight on the performance of lifelines, and domestic and non-domestic buildings in urban and small town environments at intensity MM8, and on the effect of strong shaking on previously damaged, and newly repaired, and hence weakened, structures.The good performance of reinforced concrete and retrofitted brick buildings in the Wairarapa at high intensities is of particular relevance for hazard assessment.
Discrepancies between the distribution of damaged buildings shown on a newly uncovered map of damage in Wellington CBD and the shaking zones delineated on the 1992 Wellington shaking hazard map (microzoning), and a strong correlation of damage with reclamation units, suggest the need for an intensive study of 400+ files held at the Wellington City Council Archives in conjunction with the map and newspaper accounts.
Our results show the 1942 June earthquake to be one of the more significant historical ground damaging earthquakes in New Zealand.The area in which landsliding and liquefaction ground damage occurred (about 6,500 to 11,500 km 2 ) was far more extensive than previously recognised (3,700 km 2 ) in a national reconnaissance study completed in 1997 [39].This suggests that relationships given in the 1997 study for areas affected by historical earthquakes in New Zealand are likely to be minimum values.Further studies of specific earthquakes are likely to reveal similar differences between the reconnaissance-level data and what can be learned from more in-depth research.
The distributions of landsliding and other ground damage in the June and August mainshocks are consistent with their respective isoseismal maps, for which intensities were assigned on the basis of building and contents damage rather than ground damage.The 1942 Wairarapa earthquakes provide further evidence that, after allowing for topographic and terrain effects, a zone of intense earthquake-generated landsliding can place reliable constraints on the epicentral location of an historical, or pre-historical, earthquake, the extent of fault rupture at depth, and hence, magnitude.
Although the locations of the June mainshock and distribution of its aftershocks do not define the rupture zone, they are compatible with Doser & Webb's preliminary source mechanism of a predominantly right-lateral strike-slip earthquake.Despite being shallow enough and large enough for surface rupture to have occurred, apparently it did not and the scarp-like features described by Ongley [6] in 1943 as surface fault rupture are now considered to be landsliderelated, and not tectonic in origin.This has significant implications for modelling the seismic hazard of this part of New Zealand, in that a large (Mw 7 .2),shallow (-12 km), strike-slip earthquake can occur without leaving a clear geological signal.This may lead to underestimation of the frequency of similar large damaging events.
The closeness in time and space of the earthquakes both within the sequence and with the 1934 Pahiatua earthquake, and the similarity of the sequence with the 1990 Weber earthquakes suggest that seismic hazard assessment in this part of New Zealand should take into account the possibility of similar short time-scale (within a few months) and longer time-scale (within a few years) sequences of large earthquakes in the future, together with their implications for time-variable hazard and cumulative damage.
Figure l:(a) Map of New Zealand showing the locations of seismographs in 1942, the main features of the convergent plate boundary through New Zealand (inset), and the epicentres of the 1942 Wairarapa earthquakes, all at the same location, determined by the Seismological Observatory in 1942.

Figure 1 :
Figure 1: (b) Map showing the location of places mentioned in the text.

Figure 2 :
Figure 2: Map showing the locations of instrumentally derived epicentres of the 1942 June 24, August 1 and December 2 mainshocks, June 24 foreshock, their major aftershocks and the August 1 Makuri earthquake.Also shown are major onshore faults, including the Wellington, Wairarapa, Carterton and Masterton Faults.

Figure 6 :
Figure 6: A view of Queen Street, Masterton on June 25 1942.The debris is from the second storey of Gray's brick building housing Hannah & Co., and Bullick & Blackmore.The undamaged parapet of Woolworth's single storey 1937 reinforced concrete building may be seen just beyond Gray's.(Reference No. G-123912-1/2, "Evening Post" Collection, Alexander Tumbull Library, Wellington, New Zealand).

Figure 7 :
Figure 7: Present day view of the fonner Opera House, Masterton, showing reinforced concrete beams and columns attached to brickwalls in retrofitting mostly after 1934.(Photograph by D.J.Dowrick, 1998).•

Figure 9 :
Figure 9: Present day views of 2-storey 1938 Knox Church Hall in Masterton, with a reinforced concrete frame mid brick infill, which suffered little or no damage during the 1942 earthquakes.(Photograph by D.J. Dowrick, 1998).

Figure 10 :
Figure 10: Geographic distribution of landslides and ground damage attributed to the June 24 1942 Wairarapa earthquake.Isoseismal lines are from Figure 3.

Figure 11 :
Figure 11: Sand boils at Paiaka, near the Manawatu River, caused by the 1942 June 24 Wairarapa earthquake.Note patchesin the background that also appear to be affected.

0Figure 14 :
Figure 14: Geographic distribution of landslides and ground damage attributed to the August 1 1942 Wairarapa earthquake.lsoseismal lines are from Figure 4.
n time and space of the 1942 Wairarapa earthquakes and the 1934 M,7.6 Pahiatua earthquake.The June and August !942 mainshocks are almost along strike and within 10-l 5 km of the assumed sub-surface rupture zone 153 of the predominantly right-lateral strike-slip 1934 Pahiatua earthquake [241, suggesting that stress changes caused by the I 934 event may have induced the l 942 events.Studies of stress triggering of the l 942 events (using the models of King et al [6 l], for example) wm be possible when source mechanisms of Doser & Webb are finalised, and will follow 011 from studies of other New Zealand earthquakes, such as those of Robinson & McGinty [62] and McGinty et al [63].

Figure 15 .
Figure 15.(a) (of unknown of damage to commercial structures in the Wellington CBD.Black circles show damaged structures.Structures are separated into three categories: pre-1914 (red), 1914-1934 (blue), and post 1934 (black).Non-commercial structures are shown on map but are not classified as to damage

Table 1 .
Instrumentally derived epicentres for the 1942 June 24, August I (UT) and December 2 Wairarapa earthquakes, the June 24 foreshock, the August I Makuri earthquake and major locatable aftershocks of the June 24 mainshock.The epicentres are based on New Zealand data only.The Makuri earthquake has been assigned a location based on intensity data.

Table 2 .
Summary of MM intensities for the 1942 June 24 M., 7.2 Wairarapa earthquake.Intensities of MMS and above are shown in bold.

:
Little of these costs would have been incurred if reinforced brickwork had been used.Damage to the built environment within the MMS zone, June 24 1942Many parts of the Wairarapa experienced strong ground shaking of Modified Mercalli intensity MMS or more.While no published works discuss damage in the Wairarapa, fortunately much archival material exists, particularly newspaper accounts, photographs and local authority engineers' files.

Table 3 . Summary of MM intensities for the 1942 August 1 (UT; August 2 NZT) Mw6,8 Wairarapa earthquake. Intensities of MM8 and above are shown in bold.Table 6 . Summary of MM intensities for the 1942 August 1 MwS.6 Makuri earthquake.
and in tinze had the supply to the filters cut off and the by-pass main to town in operation."In the interim the mains to town became partially emptied, the pressure fell and the artesian pumps automatically cut in to take up the supply, but in course of time all reservoirs were emptied and the Public Hospital was dependent 011 their own storage."In the meantime measures for repairs to the filters were put in hand and with the help of the Fire Brigade, the pipe trench was pumped out and new pipes, which fortunately had been kept on hand for this purpose, were placed in position.This work was completed by mid-day on Friday following the earthquake [ie about 36 hours after the main shock].... "A total of 56 water services had been repaired by 20 July.The west side of the City area seemed to have suffered most in this respect.In many cases the services were completely pulled out of the main.[Twenty five days after the earthquake, reports were still coming in of leaking services.] and before getting the main under control, a good deal of the surrounding land and the floor of the filter house was flooded.With the telepho11e out of order, it was 1101 possible for the Caretaker to obtai11 assistance, but he was ably assisted by Mrs Toms who worked with the Caretaker on the large valves, "The effect 011 our stocks of fittings may be serious on account of this unexpected call 011 supplies.Fortunately we had built up our stocks for E.P.S. [ Emergency Precautions Service] work.We will endeavour to procure further supplies as early as possible." The ruptures occur only on steep slopes and ridgelines in areas where no continuous, active, or long-lived fault trace can be seen on the airphotos.Although ground shaking features are described north to Bideford, no other ruptures are reported by Ongley, and are not evident on the photos between Bideford and the Dryers Rock fault zone .... Thus we suggest that the 1942 rupture described by