Haze, Hunger, Hesitation: Disaster aid after the 1783 Laki eruption

Abstract The 1783–1784 Laki eruption was one of the most severe natural catastrophes to occur in Iceland since settlement (around 870 CE). Vegetation damage by sulphate aerosol and fluorine poisoning caused a massive decimation of livestock which brought famine and excess deaths of 1.6 of the population. 18th Century Iceland was a Danish dependency and, despite the abundance of fish in the surrounding waters, a subsistence farming community and thus highly dependent on livestock. During the famine, the Danish government was in principle willing to provide relief. However, local authorities in Iceland were slow to ask for help, and did not dare to exploit the means at their disposal (e.g. the right to ban the export of Icelandic foodstuff) without consent from Copenhagen. The Danish officials in turn were unwilling to act decisively upon incomplete information. These two factors prevented timely measures. While 4.4 × 105kg of grain were provided for famine relief in summer 1784, the merchants exported 1.2 × 106kg of fish, which greatly aggravated the hunger in the second winter. The effects of this ‘natural’ catastrophe could therefore have been significantly reduced by efficient government measures.

Iceland was a Danish dependency and, despite the abundance of fish in the surrounding waters, a subsistence farming community and thus highly dependent on livestock. On the other hand, the farming community possessed coping strategies which mitigated the impact of livestock loss. During the famine, the Danish government was in principle willing to provide relief. However, local authorities in Iceland were slow to ask for help, and did not dare to exploit the means at their disposal (e.g. the right to ban the export of Icelandic foodstuff) without consent from Copenhagen. The Danish officials in turn were unwilling to act decisively upon incomplete information. These two factors prevented timely measures. While 4.4 × 10 5 kg of grain were provided for famine relief in summer 1784, the merchants exported 1.2 × 10 6 kg of fish, which greatly aggravated the hunger in the second winter. The effects of this 'natural' catastrophe could therefore have been significantly reduced by efficient government.

Introduction
The 1783 Laki eruption was, in terms of lava output, the second largest eruption in Iceland since the country was inhabited around AD870. Lasting from June 8th, 1783 to February 7th 1784, it produced about 15km 3 of lava, covering an area of around 600km 2 , spread fine poisonous ash over most of 5 the island, and produced a persistent sulphurous haze which was observed over large parts of the Northern hemisphere (Thordarson & Self, 1993. The so-called Haze Hardships (Móðuharðindin) caused by the eruption were probably the worst natural catastrophe which befell Iceland, and killed about 1/6 of the human population. However, few, if any, of the deaths were caused di-10 rectly by the eruption, i.e. by lava streams or tephra fall. Written accounts (e.g. Steingrímsson (1788Steingrímsson ( /1998Finnsson (1796)) agree that the main causes were famine -brought about by a massive loss of livestock -and non-specified contagious diseases (Hálfdanarson, 1984). More recently, fluorine poisoning and inhalation of volcanic haze have been suggested as additional contributors to 15 human mortality (Grattan et al., 2003;D'Alessandro, 2006;Balkanski et al., 2018). Perhaps remarkably for those times, the government of Denmark -of which Iceland was a dependency -was in principle willing to provide significant famine relief, but on the whole the operation was not successful.
sheep cows horses farms lost / given up color code depicts severity  (1983)), and livestock loss and abandoned farms (data from Rafnsson (1984a)). Livestock The haze was associated with 'sulphur dust' (deposition of sulphur compounds) and severe damage to the vegetation. Grass whitened, withered to the roots (Steingrímsson, 1788(Steingrímsson, /1998. Grass growth, and hence the hay harvest, was reduced, and it was observed that the hay had less nutritional value, and 1.3-3 120 times the normal amount was needed to feed the animals (see SM2). The situation was aggravated by the very cold winter 1783-84, which prevented grazing and inhibited grass growth in the following spring, due to frozen grounds. Other plants, including secondary food suppliers like lyme grass, berries, and Icelandic moss, were much diminished due to the eruption (Steingrímsson, 1788(Steingrímsson, /1998Pé-125 tursson et al., 1984). In many locations, fishermen did not dare to go out due to low visibility (see SM4). Many reports also note that the haze caused cold weather as it blocked the sunlight (see SM2).
Strong, low-altitude haze was observed, on and off, in all parts of Iceland during the summer and autumn of 1783 (Thordarson, 1995). Apart from the close 130 vicinity to the volcano, the northern regions (Húnavatnssýsla to Þingeyjarsýsla) report particularly strong effects, especially in mountain valleys, whereas the northwest and part of the west were affected less severely. In Isafjarðarsýsla and Hnappadalssýsla, the reduced grass growth in 1783 is attributed to hazeinduced lack of sunshine and cold, rather than poisonous fumes, which may 135 indicate less severe fog at ground level (SM2). Thin, high-altitude haze was observed above Iceland well into 1784 . In the vicinity of the lava flows, outgassing persisted through 1784, as 'the five largest clouds of smoke and steam [emerging from the lava] did not shrink at all that year' (Steingrímsson, 1788(Steingrímsson, /1998 Fine ash -produced by lava fragmentation during explosive episodes -spread over an area of 200000km 2 , including most of Iceland, except the extreme west and northwest . Although the tephra layer in most regions was not thick enough to physically damage vegetation, the fine ash carried highly toxic fluorine. It has been estimated that about 8Mt of fluorine was re-145 leased at the vents (Thordarson & Self, 1996). Fluorine can be adsorbed on fine ash particles but is subsequently washed into the ground by rainfall. Comparison with measurements from the 1970 Hekla eruption suggests a fluorine deposition by fine ash of about 9 × 10 7 kg which, when spread over an area of 200000km 2 , yields 450mg/m 2  1 . Assuming a probably optimistic 150 hay crop of 1000kg/ha (Friðriksson, 1972) yields a deposition of 3000mg/kg of hay, although instantaneous concentrations were probably less, because the fluorine was deposited over several months, and meanwhile dilution by rain water or permanent adsorption to the soil (Thorarinsson, 2012;D'Alessandro, 2006) could reduce concentrations. Still, for a sheep weighing 75kg and consuming the 155 equivalent of 3kg of dried grass/day (Sigurdarson & Pálsson, 1957), fluorine intake may have been well above 15mg/day/kg(bodyweight), which experiments by (Roholm, 1937, Ch. XXI-XXIV) suggest as an estimate for the lethal dose over periods of half a year (see also sect. 2.2.3).
Livestock was severely affected by the eruption, both through lack of fod-160 der and through fluorine poisoning. A drop in milk production to one-half or even nothing was noticed immediately after the arrival of the haze, in Vestur-Skaftárfellssýsla (Steingrímsson, 1788(Steingrímsson, /1998, but also in the North (see SM3),where the milk yield did not suffice to feed the people, let alone to set aside winter stores. Symptoms associated with fluorine poisoning of livestock (including fee-165 bleness, swellings, softened bones, loss of hair) were reported from nearly all over Iceland, except in the northwest (Pétursson et al., 1984), consistent with the spread of fine ash Thordarson, 1995) Tooth deformations called gaddur (spike) were observed from autumn 1784 onwards and could occur even years later (Finnsson, 1796). Gas poisoning or inhalation of ash 170 particles may have added to the symptoms (Pétursson et al., 1984). Animals, especially sheep, started to die within two weeks after the onset of the eruption in nearby parishes (Steingrímsson, 1788(Steingrímsson, /1998; elsewhere it took several month for livestock to die (Pétursson et al., 1984). In many regions, animals starved or had to be culled for lack of hay, and in some cases to provide meat for humans, Rafnsson (1984a) compared the number N of surviving farming animals in 1785 to 'normal' (i.e. the mean of 1703 and 1795), and found a reduction by about 1/2 for cattle and horses and 3/4 for sheep (see table 1 and fig. 2).
Apart from the impact of the eruption, this loss also reflects the impact of the 180 previous cold summer 1782, which caused lack of hay and culling of animals in the north and east (Guðjónsson, 2010, Tv_Eyjafjarðar_Dec83)], especially in the northeastern corner of Þingeyjarsýsla (Guðjónsson, 2010), and the effect of the frozen grounds and wet summer weather (especially in the south and west) in 1784, which again led to an insufficient hay harvest and further loss of 185 animals (Guðjónsson, 2010). Incomplete recovery by 1795 can lead to a, possibly regionally dependent, underestimate of the actual loss. The loss of animals was greatest close to the volcano (Vestur-and Austur-Skaftafellssýsla), but also some districts of the west (see fig. 2), while the extreme northwest and southwest were less affected. Within Árnes-and Rangárvallasýsla (Pétursson et al., 1984) and in 190 the north (Thordarson, 1995), inland communes were in general more strongly affected, which may be due to a dominantly northwestward dispersal of the ash and decreasing concentrations away from the source (Pétursson et al., 1984).

195
Wild animals, such as fresh water fish and birds, were also reduced (Steingrímsson, 1788(Steingrímsson, /1998Pétursson et al., 1984), while there is no indication that marine fish was affected.

Pollution outside Iceland and impact on climate
The transport of sulphuric haze was analysed by Thordarson (1995) (Grattan & Brayshay, 1996), but no wide-spread harvest failures (Halldórsson, 2013, p. 85 ff). Thick low-altitude haze was present, on 210 and off, until the end of July, and pulses of high intensity occurred through autumn, while the high-altitude (upper troposphere/lower stratosphere) haze remained till early 1784   , and drought in the Nile catchment (Oman et al., 2006a). Several modelling studies have investigated the dispersal and climatic effect of the Laki haze Oman et al., 2006a;Chenet et al., 2005;Oman et al., 2006b;Schmidt 220 et al., 2010;Pausata et al., 2011;Zambri et al., 2019a,b). The studies reproduce a strong sulphate aerosol haze in the northern hemisphere, both near the surface and in the upper troposphere/lower stratosphere, lasting for several months. Zambri et al. (2019a) showed that some aerosol may have reached the southern hemisphere. The haze lead to a negative radiative forcing anomaly over the 225 northern hemisphere, with peak values for late summer 1783: −5.5W/m 2 over the northern hemisphere , −4W/m 2 global mean 11 Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 25 June 2020 doi:10.20944/preprints202001.0070.v2 (Oman et al., 2006b), −12W/m 2 over the northern hemisphere, (Zambri et al., 2019b). It cooled the northern hemisphere for several months; peak values for northern hemisphere averages, October 1783: 0.35K (Highwood & Stevenson,230 2003) and 2K (Zambri et al., 2019b). The comparatively weak effect in  may be due to a low conversion of SO2 to H2SO4.
For comparison, the radiative forcing due to the the explosive Pinatubo eruption of 1991 caused a cooling of 0.4K in the global mean (Thompson et al., 2009 2019b) leading to drought in the Nile (Oman et al., 2006a), impact on cloud condensation nuclei (Schmidt et al., 2010), and stratospheric ozone concentrations (Zambri et al., 2019a). Unlike suggested by Grattan & Sadler (1999), the observed warm summer 1783 in Europe probably was a coincidence (Zambri et al., 2019b).

245
Some modelling studies also report surface concentrations of the pollutants SO2 and sulphate aerosol, although different spatial and temporal averaging methods make it difficult to compare the results. For SO2, Oman et al. (2006b) find a concentration of 10-20ppbv (25 − 50µg/m 3 ) at 65 • N , averaged over June-August and 30 • W − 45 • E, but peak concentrations can have been much higher 250 than this mean. The modelled value is above the EU health guideline (International Volcanic Health Hazard Network (IVHHN)) for annual mean SO2 concentrations (8ppbv). The guidelines also recommend that a daily mean of 48ppbv (125µg/m 2 ) should not be crossed more than three times a year. In a modelling study by Balkanski et al. (2018) poisoning have been suggested as significant contributors (D'Alessandro, 2006).
After outlining the temporal and spatial patterns of mortality, I argue that, while hunger and disease alone could explain these data, the influence of pollu-285 tion remains uncertain.   (Gunnarsson, 1983). Deaneries were roughly equivalent to the districts shown in fig. 2 A possible concern about the data is that they are based on burials, which may be delayed w.r.t to actual death. There is evidence for such delay in winter-  , 1788/1998, p.78-79), although here the delay may have been in the order of weeks rather than months. Burial delay might help to explain the sharp 'mortality' peak in June 1784in Vasey (1991 and Hálfdanarson (1984). However, [Rep1784] indicate a warm spell in November to mid-December 1783 (see SM1), which should have provided an opportunity for burials; this would mean that 330 possible delay would concern deaths after mid-December 1783. In addition, the written record from Kirkjubaejarklaustur (Steingrímsson, 1788(Steingrímsson, /1998 mentions that mortality in 1783 was not great, but surged after new year 1784; in Þingeyjarsýsla 'people began to die from hunger shortly after new year and continued to die the whole winter, spring and until now [Sept. 16th, 1784], 335 though less widely in the last two months' see SM6. Conceptions (births -9 months) were below normal from the beginning of 1784 through the first half of 1785 (Vasey, 1991), supporting the notion of severe hardships from winter 1783/84 onwards.
Investigating mortality in 1784-85, Hálfdanarson (1984  parishes, but no strong increase in June and July. Balkanski et al. (2018), using an extended data set, find an increase of 32% for June-September, without specifying monthly resolution. Grattan et al. (2003); Witham & Oppenheimer (2004) investigated English parish data accounting for 7% of the population. Both find 365 near-average mortality in June and July 1783, but a strong rise in August (127% of the normal value) and September (158%) 4 . Mortality remained somewhat elevated over the next months and peaked again in January-February 1784; this second peak might be attributed to the severe winter (Witham & Oppenheimer, 2004).

Hunger and contagious disease
In the parishes analysed by Hálfdanarson (1984), 959 deaths occurred on 1784-85. Of these, 174 were attributed to hunger and 233 to 'landfarsótt', which literally means 'land-travelling disease' and seems to have been used rather undiscriminatively for many endemic, contagious diseases, but not dis-375 tinct epidemics like smallpox (Hálfdanarson, 1984). Other causes of death include hunger-sensitive diseases like scurvy (11 cases), diarrhea (26), but also accidents, old age, and infant mortality. 341 deaths are marked as 'other and unknown'. In the north, hunger was the most frequent attributed cause of death, and landfarsótt occurred relatively scarcely. In the west, both hunger and land-380 farsótt were common, while in the northwest and southwest, only one case of starvation was recorded, but landfarsótt did occur.
Hunger as main cause of death in the north would be consistent with the fact that mortality peaked not immediately after the eruption, but in spring, when food was used up. Although the milk production dropped dramatically 385 with the arrival of the haze, Danish grain, fish, moss, and the meat of culled animals provided some food for the first few months (see SM6).
Landfarsótt raged mostly in autumn 1784 and spring 1785 and was most prevalent in the west. The connection between landfarsótt and hunger is subtle.
As Hálfdanarson (1984)  accommodation, which together with the generally poor and unhygienic housing (Magnússon, 2010, p 48ff) may have facilitated the spreading of diseases. It is therefore possible that the famine was the main driver of landfarsótt.
Regional differences in animal loss (see fig. 2) were an important, but not the only factor in determining human mortality. Animal loss, and loss of milk 415 production in summer 1783 was very high in the Skaftafellssýslur (Steingrímsson, 1788/1998), where recorded mortality was not particularly high. However, persons that fled these districts and died elsewhere do not appear in this count.
For the region closest to the eruption (Fljótshverfi, Síða, Landbrot), the death toll including fugitives was high: 224 out of 602 former inhabitants (37%) died 420 (Steingrímsson, 1788(Steingrímsson, /1998Pétursson et al., 1984). The very high death toll in The high fraction of abandoned farms in that district (see fig. 2) might partly be due to the proximity of the fishing districts which enabled migration. In Gullbringusýsla, mortality in 1785 was significantly higher than in the north-440 west (except Strandarsýsla). Both hunger and landfarsótt did occur (51 and 95 deaths, respectively, in the Hvalsnes living (Hálfdanarson, 1984)). The district received a considerable number of fugitives (see SM8), including 148 from Vestur-Skaftafellssýsla Gunnlaugsson (1984a), and 61 out of 166 deaths in Hvalsnes in 1785 were from outside the region (Hálfdanarson, 1984). Thus fugi- and had poor access to fishing.
Hunger and disease as main driver of excess mortality could explain the features of human mortality, including timing (sect. 2.2.1). This does not prove 455 that they actually were the only significant drivers. 36% of the deaths in 1784-85 in Hálfdanarson (1984 are ascribed to 'other and unknown' causes. This could be due to incomplete record keeping, but also due to distinct additional drivers of human mortality that were not recognised at that time, for example, fluorine poisoning and air pollution. The fact that the social status of 228 out 460 of 959 deaths is 'other and unknown' might suggest that record keeping was indeed not complete, but this may not explain all 'unknown' deaths.

Fluorine poisoning
Although fluorosis in humans due to volcanic eruptions is uncommon, it has been suggested that fluorine poisoning occurred after the Laki eruption as a possible disease in animals indicates that scurvy and fluorosis were easy to confuse. In the detailed descriptions by (Steingrímsson, 1788(Steingrímsson, /1998 and (Stephensen, 1785, p. 128 ff), some symptoms in humans resemble 485 symptoms found in fluorine-affected animals (e.g. loss of hair, swellings) and might thus suggest human fluorine poisoning, but can also be explained by scurvy (Roholm, 1937, Ch. XXVIII.1). Other classical scurvy symptoms like swollen gums and loss of teeth were also observed in humans (Steingrímsson, 1788(Steingrímsson, /1998 but are unrelated to fluorine poisoning (Pétursson et al., 490 1984). Scurvy would be likely to occur because of the lack of milk, an important source of vitamin C in the Icelandic diet. It may be impossible to disentangle scurvy and fluorine poisoning as they might interact physiologically, e.g. fluorine might affect the utilisation of vitamin C (Roholm, 1937, Ch. XXVII.5) whereas vitamin C helps to reduce the effects of fluorine (Gupta et al., 1996).

495
Only 11 deaths out of 959 in the data set of Hálfdanarson (1984) are listed as being caused by scurvy, most of them in Gullbringusýsla, a region with relatively low ashfall exposure, but a diet strongly based on fish, i.e. low in vitamin C.
This might simply mean that scurvy was not very wide-spread, or at least it was not (perceived as) the ultimate cause of death. Victims of fluorosis, if any, 50kg, this equals 0.75g/day. As an estimate for the lower limit of fluorine intake causing symptoms, Pratusha et al. (2011) states that 10mg/day ingested 515 for 10 years or more can cause skeletal fluorosis, and Sigurdarson & Pálsson (1957) found that 20 − 40mg/day of fluorine for 1/2 year caused mild fluorosis (slight bone changes) in sheep. As sheep have a similar body weight to humans, 20mg/day for 1/2 year will be used as an estimated threshold for mild fluorosis.
Fluorine could have been inhaled, either as HF gas or in fine ash, or ingested 520 with drinking water, meat or plant-based food. I aim to give generous upper estimates of these processes in distal areas, acknowledging that exposure close to the volcano may have been higher. HF concentration in distal volcanic plumes is generally far below health guidelines (International Volcanic Health Hazard Network (IVHHN)) and therefore seems an insignificant source of fluorine. Fine 525 ash can be inhaled while it is falling, but also when it is resuspended by wind.
After the Eyjafjallajökull eruption, Thorsteinsson et al. (2012) found PM10 concentrations of up to 1230µg/m 3 (24h mean) in nearby Vík. Ash concentrations in distal areas after the Laki eruption were probably much lower, because only a sub-millimeter ash layer was locally available for resuspension. Humans inhale 530 about 10m 3 /day of air, which would amount to 12.3mg of fine dust inhaled.
Allowing for the possibility that some particles up to 100µm could be inhaled and provide fluorine (even if intercepted in the upper airways) could quadruple the amount of dust, if the grain size distribution in Thorsteinsson et al. (2012) is representative also for Laki ash. But 50mg is still much less than the estimated 535 lethal dose of fluorine, even if the particles had consisted entirely of fluorine.
Surface water was found to contain 1-9.5ppm of fluorine after a Hekla eruption covering the area with 1-10cm of coarse tephra of a fluorine content of 70-110ppm. Laki tephra was estimated to be richer in fluorine (500ppm, Thordarson & Self (2003)), but the ash layer in distal areas was much thinner and the 540 ashfall spread over a longer time, so 10ppm (10mg/l) seems a generous upper estimate. Milk would have been safe to drink, as fluorine does not penetrate into it (Pétursson et al., 1984). The fluorine content of meat is more difficult to estimate. (Roholm, 1937, p.44) found that the ash of bones of sheep that signs of fluorosis were found in that skeleton, thus the result remains inconclusive.

Gas and aerosol
The modelling results discussed in 2.1.3 suggest that health standards for SO2 and PM2.5 (particulate matter smaller than 2.5µm) from sulphate aerosol 580 were exceeded in Iceland and probably also in Europe. Symptoms consistent with high SO2 and sulphate aerosol concentrations were reported in Iceland and Europe. Reverend Jón Steingrímsson, whose parish was closest to the volcano, mentions respiratory disorders such as difficult breathing, especially with persons suffering from pre-existing chest diseases, and irritated throats, skin and 585 eyes (Steingrímsson, 1788(Steingrímsson, /1998. Breathing problems in that region persisted at least through spring 1784, possibly due to continuing outgassing from the lava streams. However, Jón Steingrímsson explicitly states that no sudden deaths or mortal illness arose from the bad air (Steingrímsson, 1788(Steingrímsson, /1998 89) in 1783.

590
The high death rates in north Iceland (a region with high pollution exposure) have been interpreted as indication for a direct contribution of air pollution to human mortality 1783-85 (Schmidt et al., 2011). Respiratory symptoms were recorded in a document from Grund, Eyjafjarðarsýsla (cited in (Thordarson, 1995)). The þingvitni (farmers' statements) in Eyjafjarðarsýsla, Dec. 1783, re-595 ported 'disgusting stench and ill odour, such that men with [pre-existing] breast diseases temporarily stayed in bed' (see also SM5). None of these documents, nor any of the letters in [Rep1784], mention any deaths directly connected to these symptoms, or any mysterious increase in mortality in summer-autumn 1783. The overwhelming concern in the þingvitni from Eyjafjarðarsýsla is im-600 minent famine. As outlined above, greater food scarcity can also explain the high death rates in the north. Of course, contemporaries could have misdiagnosed or overlooked deaths by air pollution.
In Europe, symptoms indicating health impacts by the volcanic haze were also recorded; see Durand & Grattan (1999) and references therein. Some con- temporary sources in Europe link the haze to illness and mortality (see Grattan et al. (2003) and references therein), although in some of these sources, the wording suggests contagious disease rather than pollution, e.g. 'A fever rages in many parts, which the people term the Black Fever' (Gilpin, England, cited in Grattan et al. (2003)). As outlines above, unusual summer mortality occurred 610 in parts of England and France from August 1783. Grattan et al. (2003Grattan et al. ( , 2005 point out that excess mortality over such a large area suggests a common, probably environmental driver. Clearly, the Laki haze would be such a driver, but so would excessive summer heat. Witham & Oppenheimer (2004) Grattan et al. (2003Grattan et al. ( , 2005 all consider it puzzling that mortality only increased in August, while strong haze had been present from the end of June , and modern air pollution events (though they may be imperfect analogues to the Laki haze) affect mortality at shorter lags. For comparison, Michaud et al. (2004) found that hospital emergency room visits 625 for asthma and Chronic Obstructive Pulmonary Disease followed exposure to SO2 and fine sulphuric acid aerosol at Kilauea, Hawaii, at lags of only 1-3 days. 2009; Kochi et al., 2017) found that exposure to average concentrations of up to 45ppb SO2 over 2 years (with 100ppb exceeded 5% of the time and 5-min peaks exceeding 5ppm) were not associated with reduced lung functions, although irritations such as increased cough, throat pain and painful eyes were wide-spread. predicted 140,000 excess deaths. This is a lower mortality per population than was found for 1783 by Grattan et al. (2003Grattan et al. ( , 2005 wanted to teach man to show more caution in the future and have better control over his breadwinning' (cited in Gunnarsson (1984) making was left to lower administrative bodies, especially the Rent Chamber (Rentekammeret, the finance department), which issued orders to and received annual reports from the Icelandic officials.
There was no nobility in Iceland. The highest official on the island was the governor (stiftamtmaður); in 1783 the office was held by an elderly Norwegian 700 named Lauritz Thodal, who was regarded a competent, well-willing governor, but was suffering from ill health at the time of the eruption (Stephensen & Sigurðsson, 1854, vol., 4, p. 759 (Andrésson, 1984;Gunnarsson, 1983). For example, district commissioners had to control the quality of goods, order goods for the following year, discuss with the merchants and inform the Rent Chamber about possible complaints, and supervise emergency loans (see sect. 3.2.4).
Formally, the government system was top-down, Denmark being an abso-720 lute monarchy. In reality, Icelandic officials were often consulted by the central government (Karlsson, 2000, ch. 2.12). Still, Stephensen & Sigurðsson (1854) shows that the Danish authorities minutely regulated even minor administrative issues such as the height of contributions to an insurance for widows of Icelandic   (Gunnlaugsson, 1984b;Gunnarsson, 1983;Steingrímsson, 1791Steingrímsson, /2002 (comments by translator)).
Valuable migratory cod was abundant in the Southwest of Iceland in winter and spring, and in the Northwest in spring (i.e. in a time when there was little farm work). The cool and windy conditions were favourable for wind-drying fish. In the North and East, fishing was possible from spring (if not hindered 770 by sea-ice) to autumn, but had to be interrupted from July to mid-autumn for hay making (Ogilvie & Jónsdóttir, 2000). Farmers in these regions could send their farmhands to the southwest to either participate in the winter fisheries or barter fish for farming products.
Fishing was hampered by technological level (using open rowing boats, rather 775 than decked vessels like foreign fishermen visiting Iceland), and by administrative measures; the formation of fishing villages was prevented by a law prescribing that everybody had to be registered at a farm -either as farmer (owner or tenant) or as farmhand (Gunnarsson, 1983;Eggertsson, 1996)

Monopoly Trade
This section is based on the extensive study by Gunnarsson (1983).
Iceland was not a self-sufficient economy: It depended on the import of building wood (for houses and boats), iron (tools) and hemp (fishing lines).

800
Grain was also imported, but in normal years it was a luxury good rather than a necessity. As export goods, Iceland mainly offered wool products, sheep meat and hides, and dried fish.  Hálfdanarson (1984).
Trade was mostly carried out as barter trade, partly because the merchants refused to pay in money, hoping that this would stimulate their customers to also buy luxury goods (brandy and tobacco), which were profitable to the merchants.
However, the barter trade made it difficult for the farmers to save money for 815 bad years or even for investing in better fishing boats. All prices were fixed by the Crown. For many decades, fish prices in Iceland were very low compared to prices abroad, making dried fish a lucrative export good for the merchants, but also making the fisheries less attractive for the Icelanders. In 1855, the ban against trade with foreigners was lifted, and in the following decades the fisheries 820 expanded and helped to fuel economic growth Eggertsson (1996);Sverrisson (2002). Whether the same would have been possible in the 18th century, remains of course speculation.
The trade company was the only agency providing transport to and from Iceland, hence it was the organisation through which the government could 825 administer relief.  (Jónsson, 2009). If the traditional value ratio "1 cow = 6 ewes" reflects milk production, then milk could have provided 2500kcal/day for 57000 persons (assuming 835 625kcal/l milk). The actual population in 1703 was only 50358 (Karlsson, 2000, ch. 2.14) and had also access to other food than milk. Nonetheless, devastating famines as well as periods of local distress occurred (Finnsson, 1796). In the following, the Icelanders' possible 'lines of defence' against famine, from household to government level, are briefly discussed.

Population pressure and population control
Bishop Hannes Finnsson (Finnsson, 1796) argued that Iceland was clearly not uninhabitable, for there were enough good years to allow the population to recover from the bad ones. Yet, the population never grew much beyond 50000 in the 18th century, which has sometimes been interpreted as evidence for a 845 maximum carrying capacity of the Icelandic soil. However, Vasey (1991) argues, based on mortality data for 1740-1799, that there was no extreme pressure on food reserves in ordinary years, because mortality did not peak in spring (when food would have been scarcest), except for 6 'bad' years, including 1784 and 1785. The population ceiling was a product of birth control rather than 850 endemic hunger (Gunnarsson, 1983;Vasey, 2009): Acquiring a farm was required for marriage, and since the number of farms was roughly fixed, the number of married couples was limited. This led to high celibacy rates and late marriages.
Once married, however, Icelandic women had very high fertility rates (Vasey, 2009). Requiring a farm as base for marriage thus prevented population growth hand, the same convention helped to prevent the development of fishing villages and thus limited the access to marine resources (Gunnarsson, 1983;Eggertsson, 1996).

Food production and storage 860
The strong variability of the Icelandic climate has significant impact on grass growth and hay production. It has been estimated that a temperature anomaly of 1 degree C over October-April (i.e. a very severe winter) reduced grass growth by 30% (Bergthorsson, 1985). Both inadequate grass growth and rainy summer weather reduced the hay crop. Although most winters had mild intervals in 865 which the animals could graze, the hay harvest was of utmost importance to keep the livestock alive. In autumn, after the hay harvest, farmers had to decide how many animals they would try to keep alive over the winter; the rest was slaughtered. If the winter was colder or longer than expected, then a considerable part of the livestock could be lost. Storing extra hay was a potential 870 means to protect livestock against bad weather (or eruptions). However, farmers tended to take considerable risks, often not reducing their herds sufficiently in autumn to get their animals through a harsh winter (Eggertsson, 1998) Hardships, some would 'rather die than eating it' (Steingrímsson, 1788(Steingrímsson, /1998). Around Kirkjubaejarklaustur, several seal hunts were organised (Steingrímsson, 1788/1998). On the other hand, extreme hardships could reduce the ability to provide food; for example, people in the north were too weak to mow grass (SM2), and the district commissioner of Snaefellsnessýsla, a main fishing 895 district, complained that several boats in his district could not venture out for lack of healthy men from the north (SM4).
Marine fishing yields were, in general, only weakly correlated to farming yields (Eggertsson, 1998), thus offering an opportunity for risk spreading -provided that sufficient boats and tools were present for additional fishing efforts.

900
After the Laki eruption, the fish itself probably didn't suffer. In summer and autumn 1783, poor visibility hampered fishing in many regions (see SM4), although in Eyjafjarðarsýsla there were farmers who switched to fishing to make up for the lack of milk (see SM2). While especially the north was plagued with sea ice in the first half of 1784, elsewhere fish catches were mixed (see SM4).

905
In the main fishing regions in the west and southwest, the spring fishing was not bad overall, and the Danish merchants were able to buy roughly the average amount of fish in winter/spring 1784 (Andrésson, 1984

Distribution and social safety net
At least in principle, everybody in Iceland was entitled to food and shelter (Eggertsson, 1998). Servants were mostly paid in food, shelter and clothes, and working contracts between farmers and farmhands started in the beginning of 920 June (i.e. the busy farming season) and lasted for a year; servants could thus not be turned out in winter or in case of temporary illness. Households with sufficient means were obliged to take in their poor relatives, and poor people without suitable relatives had to provided for by their commune (hreppur). The communes could also support households that were in temporary difficulties.

925
Food was thus fairly well distributed among farmers, workers, and paupers on communal assistance, although vagrants and beggars also existed. However, this safety net operated only on a local level (there were about 160 communes in Iceland), hence climate or volcanic risks could affect the whole commune and overwhelm the system. After the Laki eruption, several communes had more 930 than two paupers per household, partly because farmers could not afford to hire farmhands, who then became paupers (Finnsson, 1796). In many regions, farmers were forced to give up their farms and become vagrants (see also SM8).
There was no strong relief organisation beyond the local level, except some church charity. The bishops of Skálholt donated 20 ríkisdalir from a 'fund for 935 the poor' to the parishes closest to the Lakagígar (Steingrímsson, 1791(Steingrímsson, /2002) -the value of 20 ewes for a population of several hundred persons. Neither the communes, nor another authority, organised food stores; this would likely have been difficult because building and transport were expensive. The Land Commission of 1770 suggested to build emergency stores at all trading centres to 940 prevent famine. This suggestion was not carried out (Andrésson, 1984); instead, an emergency credit system was decided upon, see sect. 3.2.4.

Trade
Trade can smoothen local food shortages. Domestic trade within Iceland was common, e.g. farming against fishing products. However, under severe distress, 945 it could happen that farmers (or fishermen) had nothing to barter. Transport was also a limiting factor, because overland transport relied on horses (and healthy men), which often lacked during farming crises (see SM7&8). Coastal shipping or navigable rivers hardly existed.
The monopoly trade company both imported and exported food. Grain im-950 port averaged 16950 Danish tons or 1.4 × 10 6 kg over 1763-84 (Andrésson, 1984), whereas the export of mutton amounted to 3223 barrels (4 × 10 5 kg) around 1770 and dried fish to 8120 skippund (1.3 × 10 6 kg) (Gunnarsson, 1983). As a rough estimate, assuming a population of 50000 and a daily calorie intake of 2500kcal/person, imports and exports amounted to 41 and 47 daily rations, re- obtaining goods for credit in early summer and paying with farming products by the end of the year (Gunnarsson, 1983).
The trade regulations from 1776 stipulated that if widespread hunger threatened, the governor, deputy governor and district commissioners could ban the export of Icelandic foodstuff (Andrésson, 1984). In addition, the merchants were 970 obliged, in cooperation with the district commissioners and communal overseers, to give farmers in distress an emergency loan of foodstuff and tools, typically for one year. This rule was clearly meant as a temporary relief measure and did apply neither to officials, who were considered wealthy enough to not need emergency loans, nor to persons considered unlikely to be able to pay back the loan. of food in Iceland, particularly in winter. To make things worse, in early 1783, the authorities in Copenhagen felt that the emergency loans had been abused by persons not deserving them, including officials, and sent stern orders both to 980 the merchants, who should collect outstanding debts and give fewer credits, and to the Icelandic officials to pay their debts and be less generous in suggesting ordinary farmers for loans (Andrésson, 1984).
Increasing food import during a crisis was time-consuming, because shipping between Iceland and Denmark only occurred in summer, i.e. orders could be 985 placed only for next year. In 1774 and 1778, it was discussed whether a Danish ship should annually be sent to Reykjavík in autumn, stay there over winter and return with news in spring (Stephensen & Sigurðsson, 1854, vol. 4, p.107, p.437). These texts do not say whether this was actually done; at least in spring 1784, no ship was sent from Reykjavík to Denmark. In 1787, this postal service 990 was definitely established (Stephensen & Sigurðsson, 1854, vol. 5, p. 432).

Summary
In normal years, food in Iceland was not scarce. Although Iceland's economy in the 18th century has been described as poor, stagnant, and under-exploiting its resources, especially fish (Gunnarsson, 1980;Eggertsson, 1996), the farming 995 society had buffers on the farm and commune level to weather less severe drops in food supply. These buffers obviously helped to reduce mortality during large crises, because the loss of human lives 1784-85 was much less than the loss of farming animals. Tying marriage to acquiring a farm provided an effective birth control and ensured that population did not overstretch the carrying capacity  (Rafnsson, 1984b   15th (Stephensen, 1783). The governor, Lauritz Thodal, only informed the government on September 16th (Gunnlaugsson & Rafnsson, 1984, comment by editors in section II), because he first wanted to gather more precise information -a serious delay seeing that autumn was approaching and thus the time window for shipping over the North Atlantic was closing. The first news of the eruption

The investigation ship and Danish hibernation
Despite the vagueness of the first reports, the Rent Chamber took action and decided on September 17th to send a ship to Iceland to investigate the situation (Gunnlaugsson, 1984b). The ship was loaded with some grain, and aboard were two emissaries, the young lord-in-waiting and Rent Chamber member Hans 1050 C.D.V. von Levetzow "who likely desired soon to take the place of the current governor of Iceland" (Stephensen, 1888, p.229)  Reykjanes during a submarine eruption. The Danish authorities were anxious to take formal possession of this island to forestall other nations to use it as a base for fishing in Icelandic waters or even break the trade monopoly (Stephensen & Sigurðsson, 1854, vol. 4, p.744ff). They needn't have worried: The island was eroded before anyone found it again (Stephensen, 1888, p.252). The ship departed another three and a half weeks after the Rent Chamber session, on October 11th. Around that time, further worrying news arrived from Iceland, including Thodal's report and Jón Steingrímsson's letter. The Danish authorities were now convinced that something serious was happening in southern Iceland (but had no idea that the north might also be affected).
Meanwhile, the investigation ship had run into several Atlantic autumn 1080 storms. After three attempts to reach Iceland, it took winter shelter near Kristianssand in southern Norway (Stephensen, 1888, p.236-237 mercy, and the merchants were reluctant to disregard the recent letters without consent from Copenhagen (Andrésson, 1984). This consent, of course, could not be obtained with winter approaching.
Similar difficulties arose concerning the ban of exporting Icelandic foodstuff (Andrésson, 1984). The merchants had direct financial interests to export as 1110 much as possible from Iceland, because merchants received 1.5% of the value of their exports as top-up on wages and merchants' assistants 0.5%. In late summer 1783, the Icelandic governor and district commissioners did not enforce an export ban. Of course, the crisis had not fully unfolded by then, but the withering of the vegetation and reduced milk production had manifested themselves 1115 in many trade districts. As a result of the exports, no significant emergency stores were at hand in Iceland in autumn 1783.
In the course of the winter and spring 1784, the fishing season in southwestern and west Iceland was not bad, and the merchants succeeded in acquiring the usual amount of fish from Icelandic fishing boats: around 1.5 million kg 1120 (Andrésson, 1984). In addition, the trade company caught fish with its own vessels. In spring 1784, Thodal banned the export of butter and tallow, but not fish (probably the most desired export good). Stefán Þórarinsson banned the export of all Icelandic foodstuff from his harbours until further notice; the merchants were obliged to sell such goods back to the population for the pur- fishing and, due to the livestock decimation, also very limited meat products, hence this export ban was little effective.
On the district level, while district commissioners complain about the unwillingness of merchants to give loans, relatively few report having tried to confront 1130 merchants or having decreed export bans (see SM7). The district commissioner of Suður-Múlasýsla had forbidden the export of meat and tallow in July 1784, but could only express his hope to the Rent Chamber that the merchants would be held responsible in case they disobeyed. His colleague in Norður-Múlasýsla complained that district commissioners had no legal means against merchants 1135 (except sending a complaint to Copenhagen). In north Iceland, 4400kg of dried fish were exported despite the ban (Andrésson, 1984). But there are also examples of compliant merchants who willingly handed out foodstuff, e.g. in Skagafjarðarsýsla (see SM7), and the district commissioner of Rangárvallasýsla organised grain to be handed out to fugitives from Vestur-Skaftafellssýsla (Ste- or Thodal ignored his letter; but certainly Thodal did neither send Icelandic fish to the northern harbours nor decree a full ban on exporting Icelandic foodstuff (Andrésson, 1984).

Flour and fish (summer 1784)
In spring 1784, the ordinary trading ships were sent to Iceland earlier than 1175 usual and given strict orders to do everything possible to reach their destination (Andrésson, 1984). Should a harbour be blocked by sea ice, the ship should not return to Denmark but wait in the vicinity for the ice to break. However, no significant additional amount of foodstuff was shipped to Iceland this spring: Compared to the 1764-1784 mean of 16950 tons, 24203 tons of grain were im-1180 ported in 1784, i.e. 7073 tons above average; but of these, 5300 tons were only shipped after mid-July.
On April 19th, the Crown issued a decree (Gunnlaugsson, 1984b;Stephensen & Sigurðsson, 1854, vol. 5, p. 45-46) that Thodal and Levetzow, together with the local administrators (district commissioners, deans and parsons), should 1185 collect information about which inhabitants were in need of food aid, and how help could be administered. It was decreed that no food aid was to be handed out unless under the supervision of Thodal or Levetzow. Local administrators were ordered to help with collecting information concerning the needs of the victims of the eruptions for aid (building material, animals, food) and the pos-1190 sibilities to relocate fugitives. The Danish authorities clearly still believed that the catastrophe was only regional, in particular, it did not occur to them that northern Iceland (the region hardest hit by the famine) might be affected at all (Gunnlaugsson, 1984b). Also, the Danish officials apparently considered it essential to collect all possible data on the calamity prior to spending money on ríkisdalir on his own account. This act of disobedience brought about a lawsuit against the dean, although eventually he was condemned only to a minor fine of 1205 five ríkisdalir and a public apology (Steingrímsson, 1791/2002. As mentioned, the news of the devastating famine reached Copenhagen in mid-July 1784. Now that the government had received certainty about the situation, swift action was taken to meet the emergency. On July 21st, it was decided to send 3000-4000 Danish tons of flour to Iceland (Gunnlaugsson, 1984b;1210 Stephensen & Sigurðsson, 1854, vol. 5, p.99-100); by autumn, 5300 tons had been sent (Stephensen & Sigurðsson, 1854, vol. 5, p.106-107). In addition, a letter dated July 17th was sent to Eyrarbakki, decreeing that part of the fish catch from West Iceland (which was of lower quality than the fish of Southwest Ice- where the need for food was greatest, i.e. northern and eastern Iceland. Unfortunately, only one of the ships sent to Iceland carried instructions concerning the fish, and this ship was shipwrecked off the coast of Vestur-Skaftafellssýsla, and the letters got lost (Andrésson, 1984). Governor Thodal and the district commissioners in the fishing regions still did not dare to declare a ban on ex-1220 porting foodstuff without explicit orders from Denmark. Thus in late summer of 1784, the merchants exported nearly all fish they had acquired during the last spring, namely 7558 skippund (=1,200,000kg) bought from Icelanders plus their own catches. No fish transports took place towards the north and east (Andrésson, 1984). In the following winter, another several thousand Icelanders 1225 died, for a large part of starvation (or landfarsótt). 1,200,000kg of fish could have provided 2500kcal/day to 50000 persons for about 5 weeks. Meanwhile, the Danish merchants profited considerably less from the fish than expected: Fish prices, which had been unusually high during the American war of Independence, had dropped dramatically after the Treaty of Paris in summer 1783, 1230 from 0.17 ríkisdalir/kg (averaged over 1780-82) to 0.12 ríkisdalir/kg (1783-87) (Gunnarsson, 1983, p. 151).
Even the food aid which did reach Iceland was not necessarily effective, especially in the remote areas, due to the lack of horses required to transport food from the harbours overland (Andrésson (1984) SM7). An attempt to ship 1235 some grain from the Vestmannaeyjar trading post to Dyrhólaey (see fig. 2) in the particularly remote Vestur-Skaftafellssýsla district was given up due to bad weather. Reverend Jón Steingrímsson complained that it would have been more effective to provide his parishioners with fishing and sealing gear, which would have allowed them to feed themselves to some extent (Steingrímsson, 1791(Steingrímsson, /2002(Steingrímsson, , 1240 p. 84-85). hundred farms and about 10 churches had been severely damaged or even completely collapsed (Guðjónsson, 2010).
It has long been claimed in Icelandic history books that the Danish authori-1250 ties now considered Iceland uninhabitable and contemplated a complete evacuation by relocating the remaining Icelandic population to Jutland in Denmark.
However, written protocols do not support this hypothesis, although it remains possible that a complete or at least large relocation has been considered orally and informally (Gunnlaugsson, 1984b). What has been considered officially is population as a breech of law and pointed out that Iceland had no resources to feed hungry soldiers (Eirìksson, 1984). The whole plan was given up shortly afterwards.

Donations and Debts (1785 and beyond)
In February 1785, a special commission (named 'the later land commission', 1265 landsnefndin síðari) was set up to investigate how to restore the Icelandic economy (Gunnlaugsson, 1984b;Stephensen & Sigurðsson, 1854, vol. 5, p. 118-120;124-127). Among its members were Jón Eiríksson of the Rent Chamber, the executive board or the trade company, and Levetzow, who by then was appointed to replace the retiring Thodal as governor in April. The commission decided to put an end to the aforementioned evacuation plans (Stephensen & Sigurðsson, 1854, vol. 5, p. 216 ff), and to hold a second collection of money, this time in all market towns of Denmark-Norway (Stephensen & Sigurðsson, 1854, vol. 5, p.123-124). This took several months to organise, but eventually, about 36000 ríkisdalir were collected in 1785 (Gun-1275(Gun- narsson, 1983. Some further aid measures were decided upon and confirmed by royal decree on June 22nd, 1785 (Gunnlaugsson, 1984b;Stephensen & Sigurðsson, 1854, vol. 5, p. 216  bakki and handed out to the victims of the earthquake who needed to rebuilt their homes (Stephensen & Sigurðsson, 1854, vol. 5, p. 121-123 ff). Timber and further material for building boats were also to be sent to the fishing districts of Gullbringusýsla and Snaefellssýsla, such that the fugitives from the North and Southeast could settle down as fishermen. Governor Levetzow had to supervise 1290 the handing out of the timber. Finally, the Rent Chamber sent orders that spring to the district commissioners to count the population and the remaining livestock. In particular, it should be investigated which farms were in urgent need of additional livestock to remain inhabitable; it was planned to provide these farms with money (from the collection funds) to acquire animals.

1295
Not all of these measured proved as effective as was hoped. The Danes sent almost twice as much grain as usual (32200 rather than 16950 Danish tons), and three (not four) shiploads of fish were sent to the northern and eastern harbours, but no other shipments in between harbours were made (Andrésson, 1984), and there were repeated complaints by the Icelanders that it was difficult to obtain 1984b). Note also that the food aid was not a gift; it was sold (albeit without profit) or handed out as emergency loan to eligible people, i.e. farmers in acute distress who were however expected to pay their debts.
Concerning the timber, it appears that Levetzow was very hesitant to hand 1305 it out (even though it was already paid for by the collection money), setting up a complicated bureaucracy for the applicants to prove their need (Andrésson, 1984). At the end, a good part of the timber, both in Eyrarbakki and the fishing districts, was never handed out but remained in the merchants' store and was sold in the course of time as ordinary merchandise. About the boat timber,

1310
Levetzow claimed that no new boats were needed because enough people had died the last two years to free up boat places for the fugitives, though it may be that he acted to please incumbent boat owners, who feared that new boats would make it more difficult for them to find labourers for their own vessels (Andrésson, 1984).

1315
In some cases, fugitives were also sent back. For example, 40 paupers who had fled westward from eastern Vestur-Skaftafellssýla, were forced by Levetzow to return to their homes in early autumn 1785. By law, paupers were entitled to poor relief in their home commune. However, the 90 remaining, impoverished inhabitants had no means to provide for the 40 returning fugitives, and even 1320 the charitable parson Jón Steingrímsson wrote that nothing could be done but 'simply finding them a place to die' (Steingrímsson, 1788(Steingrímsson, /1998. At the end, the parish was saved by an exceptional catch of seals, but the episode clearly illustrates how rigidly authorities applied the law, and that aid was far from sufficient.
These orders were given despite the significant amount of money gathered in the collection in 1785. The collected money remained largely unused and was 1335 saved as the so-called 'collection funds' in case Iceland should ever be hit by hardships again. The funds depreciated due to inflation and was eventually used in the 1840ies to construct a high school building in Reykjavík (Gunnarsson, 1983, p.145-146).
The attempt to aid farmers in urgent need for animals to replenish their 1340 livestock was not very successful: Not only was gathering the information a slow process, but what was worse, animals were scarce in the whole country, and it was nearly impossible to buy them (Gunnlaugsson, 1984b). A Danish request to England in February 1785 to export some English sheep to Iceland had been refused as the English were keen to protect their wool export (Agnarsdóttir, 1345 1992), and the Danes did not pursue the matter (Stephensen & Sigurðsson, 1854, vol. 5, p. 216 ff). Still, some money (from the collection funds) was handed out to farmers in the following years and may have been of some help, although, as Jón Steingrímsson remarked, 'A great number of farmers and farms could have been restored more quickly if the money, which was given to them 1350 for the purchase of livestock, had not been taken back for the payment of rents and other debts' (Steingrímsson, 1788(Steingrímsson, /1998. The considerable amount of unpaid Icelandic debts with the trade company 1783-88 may have been due not to leniency, but to the fact that many debtors had died of hunger (Gunnarsson, 1984).

Recovery and lasting effects of the Haze Hardships
In his treatise on famines in Iceland, Finnsson (1796) wrote that 'Iceland never has been defeated by bad years to that extent, that during better years it could not recover, and feed its children'. This also holds for the Laki famine.
Livestock was nearly restored after 12 years (see table 1). The population began than died each single year from 1787 to 1801, and in the decade 1791-1801 the population increase was 1-2% each year (Gunnarsson, 1983, fig. 2.1). This rapid increase was facilitated by the high fertility of married Icelandic women (Vasey, 2009), and by the requirement of farm possession for marriage, which created 1365 a 'reserve' of unmarried labourers who could take over the deserted farms and found a family. In the region near Kirkjubaejarklaustur, 43 out of 47 destroyed farms and 9 out of 14 deserted crofts were eventually rebuilt, in some cases after re-location.
The events of 1783-85 made a small contribution towards urbanisation in   (Gunnarsson, 1983, p.148-149 determined the vulnerability to famine (sect. 6.1). Next, the magnitude of the disaster aid expenditure (sect. 6.2) will be discussed, and finally, the way in which these resources were put to use (sect. 6.3).
6.1. Development aid without structural change 1425 Were the Danes to blame for Icelandic poverty and the catastrophic impact of climatic and volcanic events in the 18th century, as the 19th century nationalists claimed (Gunnarsson, 1984;Oslund, 2011, ch.  list to the disfavour of the Icelanders, such as to be able to extract higher harbour rents from the merchants; his successor had to revert the change in 1703 after several years of famine in Iceland (Gunnarsson, 1983, p. 55-56 subsistence farming system rather than structural change. Attempts to stimulate fishing were almost bound to be unsuccessful due to the laws restricting free labour; inconsistently, these laws were sharpened rather than abolished in early 1783 under the pressure of the landowning elite Gunnarsson (1983).

1455
Maybe the optimistic officials underestimated the difficulties of 'developing' a backward economy from behind a clerk's desk, with insufficient understanding of the societal fabric. The modern aim of (economic) 'progress' was only beginning to emerge in Europe (Ferguson, 2018, chapter 2)  the Crown (ca. 260000 ríkisdalir) and private shareholders (Gunnarsson, 1983, p.142,144). It has been argued that these losses can partly be seen as indirect aid (e.g. unpaid Icelandic debts), while a substantial part of these losses was also caused by changing market prices outside Iceland and liquidation of the company 1787-88 (Gunnarsson, 1983, p.146  share holders. When these companies went more or less bankrupt after the end of the American war of Independence, the Crown decided to compensate the shareholders for their lost capital and non-forthcoming profits by paying them 7.8 million ríkisdalir over the next years, 100 times as much as the direct aid for Iceland (Gunnarsson, 1983, p.141). Another, albeit trifling, expenditure 1500 may illustrate royal priorities. In normal years, the Crown imported 50 falcons from Iceland, but in early 1785 it was feared that no oxen could be purchased there to feed the falcons during the journey to Denmark. It was thus decided to limit the import to 30 falcons and send 20 living oxen to Iceland, to feed not the starving Icelanders but the royal falcons. The additional costs (including 1505 rebuilding the falcon ship to transport the oxen) were estimated to be 1896 ríkisdalir (Stephensen & Sigurðsson, 1854, vol. 5, p. 128-129). These two examples suggest that, given the will, the Danish Crown could have afforded to spend more to save its Icelandic subjects from starvation.
The Danish reaction the Laki eruption has been criticised both by contem- narsson, 1984; Oslund, 2011, Ch. 1). However, large-scale government relief was by no means the obvious reaction to famine in earlier centuries (Gunnarsson, 1980); this not only holds before the Laki eruption, but also 65 years later, 1515 e.g. during the Irish potato famine . During that episode, initial (costly but insufficient) relief schemes were gradually abandoned for fear of disturbing the market. Unlike in Iceland 1783-84, this cannot be explained by unreliable transport and troubled communication over a stormy North Atlantic, but rather by a lack of political will: In the 1840ies the 'Laissez-faire' ideology had much 1520 influence on policy and government interference was considered harmful since 'if left to the natural law of distribution, those who deserve more would obtain it' (Ó Gráda, 2000, p.6-7). The Laki eruption, on the other hand, took place during a relatively enlightened period in which it was considered good governance to actively foster the economic activity of a country's subjects and to 1525 mitigate famine (Gunnarsson, 1984(Gunnarsson, , 1980. Thus, however insufficient, belated and clumsy the Danish aid may have been, one should acknowledge that they at least tried to help.

A case study in disaster (mis)management?
Why was the disaster relief not more successful? It is interesting to dis-1530 cuss this question from a disaster risk reduction perspective, even though this is 'unfair' in the sense that disaster risk reduction is a fairly modern concept.
It should also be acknowledged that the Danish authorities worked under severe logistic constraints. Iceland was a remote dependency which most Rent Chamber members were not familiar with, although one high-placed member,

1535
Jón Eiríksson, was a native Icelander. Communication and transport to Iceland were cumbersome and expensive, the duration of a single journey was of the order of 2-3 weeks, and in winter sailing was (considered) impossible. Transport within Iceland was likewise difficult, especially after the massive loss of horses.
The first step to take measures is to detect the threatening disaster and raise the island). However, when famine loomed, there seems to have been confusion as to what was an 'emergency' and who was 'eligible' for a loan. The Danish government had further undermined its own emergency plan in early 1783 by sending letters concerning the need to reduce Icelandic debts, underestimating the tendency of Icelandic authorities and trade representatives to follow the most 1595 recent instructions rather than the overarching goal of preventing starvation.
As described in sect. 6.2, the Danish government formulated in autumn 1783 a clarification that loans were still to be given in emergencies, but only after the investigation ship had departed.
Another difficulty diminishing the efficiency of aid were conflicts of inter-1600 ests. The most striking one is the multitude of roles of the trade company.
Being the only organisation providing transport to Iceland, and the only owner of significant food stores on the island (at least between the fishing season in spring and the departure of the trade ships in summer), it was the instrument 60 Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 25 June 2020 doi:10.20944/preprints202001.0070.v2 through which the government could administer relief. At the same time, the 1605 trade company was a commercial enterprise, and both the shareholders (including the Crown) and the employees in Iceland expected to make profit from it.
On the Copenhagen end, the Crown could, if it wished, override the shareholders' economic interests and decree that unprofitable rescue actions be carried out. However, the trade representatives in Iceland had a strong financial incen-1610 tive to export as much from the island as they could, and thus to oppose any attempt by Icelandic officials to ban export. They also refused in some cases to put their large fishing vessels at the disposal of the governor for transporting foodstuff, possibly because they considered fishing more profitable. Maybe the central government did not foresee this problem; at least no reference regarding 1615 financial compensation is made in the order of July 21st, 1784 (Stephensen & Sigurðsson, 1854, vol. 5, p.99-100), which (unsuccessfully) ordered that local tradesmen should 'provide their Hukkerter [fishing vessels] to transport fish and other foodstuff from one district to the other' (see sect. 4.2.3).
Finally, while the Danish government was willing to take significant action 'These [census lists] could hardly be expected to make sense or to agree, as people were constantly moving back and forth and some dying' (Steingrímsson, 1788(Steingrímsson, /1998. In any way, collecting information in the large, thinly populated Iceland was a tedious business, and the data could be shipped to Copenhagen only with the ships departing in autumn, so that they would be 1640 acted upon only in the next year. In spring 1785, renewed requests for a careful survey of livestock and human population were sent to the whole of Iceland, partly to assess which farms were in need of assistance to buy livestock. From many districts, this information was delivered only in 1786. In short, it seems that the Danish government was so afraid to incur aid expenditures that might 1645 later prove unnecessary, that it preferred to delay action by a year or more and risk that the aid might come too late to do any good. If one tries to find one single expression describing the shortcoming of the government response, it is undue optimism. The Danish officials hoped that the effects of the eruption would not be too bad, trusted that information from and 1650 its own orders to Iceland would be transmitted smoothly, that all (sometimes unclear) orders would be carried out immediately and effectively, with officials and trade representatives functioning perfectly without frictions such as competing interests. Maybe a more efficient aid could have been accomplished if the officials had constantly asked themselves: How can this measure go wrong

1655
-and what can be done to mitigate potential failure? But this would have required much foresight, imagination, and an intimate knowledge of the local geographical and societal situation. Funding: