Hekla 1947, 1845, 1510 and 1158 tephra in Finland: challenges of tracing tephra from moderate eruptions

Several cryptotephra layers that originate from Icelandic volcanic eruptions with a volcanic explosivity index (VEI) of ≤ 4 and tephra volumes of < 1 km3 have previously been identified in Northern Europe, albeit within a restricted geographical area. One of these is the Hekla 1947 tephra that formed a visible fall‐out in southern Finland. We searched for the Hekla 1947 tephra from peat archives within the previously inferred fall‐out zone but found no evidence of its presence. Instead, we report the first identification of Hekla 1845 and Hekla 1510 cryptotephra layers outside of Iceland, the Faroe Islands, Ireland and the UK. Additionally, Hekla 1158 tephra was found in Finland for the first time. Our results confirm that Icelandic eruptions of moderate size can form cryptotephra deposits that are extensive enough to be used in inter‐regional correlations of environmental archives and carry a great potential for refining regional tephrochronological frameworks. Our results also reveal that Icelandic tephra has been dispersed into Finnish airspace at least seven times during the past millennium and in addition to a direct eastward route the ash clouds can travel either via a northerly or a southerly transport pathway.


Introduction
Cryptotephra layers produced by very large (VEI ≥ 5: Newhall and Self, 1982) silicic explosive eruptions of Icelandic volcanoes form the backbone of the North European tephrochronological frameworks and provide powerful tools for dating and correlating palaeoenvironmental archives such as peat sequences and lake sediment records from the Last Glacial-Interglacial transition (LGIT) and the Holocene (e.g. Wastegård and Davies, 2009;Lawson et al., 2012;Davies et al., 2012;Timms et al., 2019). In addition to these major marker layers, cryptotephra horizons from Icelandic smaller scale (VEI ≤ 4) eruptions (e.g. Dugmore et al., 1996;Rea et al., 2012;Watson et al., 2015), as well as ultra-distal cryptotephra deposits that are sourced from other continents and volcanic centres, have been detected across Northern Europe (e.g. Jensen et al., 2014;van der Bilt et al., 2017;Plunkett and Pilcher, 2018;Jones et al., 2019). Icelandic eruptions of VEI ≤ 4 have generally produced tephra isochrones with restricted dispersal areas in the distal field (Lawson et al., 2012) and most of the ultra-distal cryptotephras have been identified at single sites thus far (e.g. Plunkett and Pilcher, 2018;van der Bilt and Lane, 2019). However, the potential for both ultra-distal tephra and tephra from Icelandic smaller scale eruptions to be used as regional and inter-regional isochrones may improve with new and better methods for extracting and analysing small amounts of tephra from peat and minerogenic deposits. For example, picking individual glass shards for geochemical analysis using a micromanipulator (Lane et al., 2014;MacLeod et al., 2014) and optimising the electron microprobe analysis protocols for geochemical fingerprinting of single glass grains as small as 10 µm (Hayward, 2012) enable source eruptions of cryptotephra deposits consisting of very scarce and small shards to be traced.
So far, the majority of the historical (younger than 870 AD) tephras originating from Icelandic smaller scale eruptions remain undetected in distal areas. However, a simulation of transport, dispersion and deposition of volcanic ash from an explosive Hekla eruption based on an atmospheric circulation model demonstrated that an initial plume height of just 12 km would give a > 45% probability of ash fall-out in Scotland and Fennoscandia (Leadbetter and Hort, 2011). As all the 18 historical Hekla eruptions have started with an explosive phase during which a sustained tephra plume has reached a height of 12-36 km (Thorarinsson, 1967;Janebo et al., 2016), each of these eruptions, independent of size, carries the potential to form distally transported cryptotephra deposits in the North European palaeoenvironmental records.
The scarcity of cryptotephra deposits from historical Hekla eruptions could partly be explained by the fact that most of the tephrochronological work that uses the recently developed methods of concentrating scarce shards has focused on LGIT tephras (e.g. Larsen and Noe-Nygaard, 2013;MacLeod et al., 2014;Jones et al., 2018). However, the occurrence of diffuse cryptotephra layers or deposits with insufficient shard concentration for geochemical analysis is relatively common at shallow depths in peat stratigraphies (e.g. Boygle, 2004;Wastegård et al., 2008;Housley et al., 2010;Watson et al., 2015) and could indicate the presence of tephra from several historical Icelandic eruptions. Geochemically identified cryptotephra horizons produced by eruptions of VEI ≤ 4 include three separate historical Hekla layers that are present in Ireland and the UK; namely, Hekla 1510 (Dugmore et al., 1996;Pilcher et al., 1996;Reilly and Mitchell, 2015), Hekla 1845 (Watson et al., 2015(Watson et al., , 2017 and Hekla 1947 (Rea et al., 2012;Watson et al., 2015). Each of these layers has been used as a local or regional correlation and dating horizon in environmental research in the UK and Ireland (Hall, 1998;Housley et al., 2010;Watson et al., 2015), but it is not yet well established whether these smaller scale tephras could form isochrones that are extensive enough to allow for inter-regional correlations in the distal area. Until now, no distal deposits from the Hekla 1947, 1845 or 1510 eruptions have been identified with certainty further afield outside of Ireland and the UK, even if contemporary eyewitness accounts reveal that the Hekla 1845 and 1947 tephras formed visible fall-out in the Shetland Islands and Finland, respectively (Salmi 1948;Thorarinsson, 1981). Additionally, a cryptotephra layer in the Faroe Islands has been tentatively correlated with the Hekla 1845 eruption (Wastegård, 2002), and a diffuse deposit consisting of very scarce shards that possibly originate from historical eruption (s) of Hekla was recently reported from northwestern Russia (Vakhrameeva et al., 2020).
In this paper we discuss several attempts at finding the Hekla 1947 tephra as well as other Icelandic historical tephras from 25 Finnish peatlands. We report the first identification of the Hekla 1845, Hekla 1510 and Hekla 1158 cryptotephra deposits in southern and central Finland, thus extending their known dispersal areas significantly further east. We also present new results on the geochemical composition of Hekla 1845 tephra from a proximal site in Iceland as well as new data on the physical properties and geochemical composition of the Hekla 1947 tephra sample collection (Salmi, 1948). Finally, we discuss preservation issues of scarce cryptotephra deposits in environmental archives and the potential for smaller scale eruptions to form widespread isochrones in the far-distal field.

Site selection and sampling
Our main research area comprises 18 peatlands (Table 1, Fig. 1) that are within the fall-out zone of the Hekla 1947 tephra in southern Finland as inferred by Salmi (1948).
Additionally, four peatland sites outside of the fall-out zone (7, 15, 18 and 20 in Fig. 1) and three sites in west-central Finland, in the proximity of an unconfirmed eyewitness report of tephra fall-out (Salmi 1948) were chosen for investigation (10-12 in Fig. 1). A previous search for Hekla 1947 tephra at bog and lake sites in Finland failed to detect any shards within the fall-out zone (Kalliokoski et al., 2019), which could indicate either a highly patchy tephra fall-out within the dispersal area, a sporadic occurrence of the tephra in environmental archives due to post-depositional processes, or preservation issues (Boygle, 1999;Bergman et al., 2004;Payne and Gehrels, 2010;Watson et al., 2015). Given that a possible connection between precipitation and the amount of tephra fall-out has been suggested in earlier studies (Dugmore et al., 1995;Langdon and Barber, 2004;Rea et al., 2012), precipitation maps for the days following the 29 March 1947 Hekla eruption were generated from the Finnish Meteorological Office weather data ( Fig. 2A-C) and used as an aid for site selection. Sites with both high and low precipitation were selected ( Fig. 1 and 2A-C) to test for a correlation between shard concentration and precipitation amount. Additionally, a snow-depth map for 30 March 1947 was prepared from the Finnish Meteorological Office data ( Fig. 2D) to search for potentially snow-free peatlands where spring melt would not have redistributed the tephra deposits (Boygle, 1999;Bergman et al., 2004). Peatlands within a 20 km radius of the nearest weather station were targeted to ensure the validity of weather data at our research sites (Fig. 1). To minimise the likelihood of post-depositional disturbance of tephra layers by anthropogenic activity, peatlands in nature conservation areas were prioritised. Therefore, protected blanket mires were preferred over drained or cut ombrotrophic peat bogs, even if it has been suggested that raised peat bogs better record and preserve primary atmospheric fall-out of tephra (e.g. Persson, 1966;Bergman et al., 2004;Gehrels et al., 2008).
Peat samples were collected from each peatland site during the summer field seasons of 2014 and 2015 using a Russian peat corer with a 50 cm long and 5 cm wide cylinder. Successive samples with a 10 cm overlap were cored from the full peat stratigraphy. An additional 20-50 cm long monolith with a cross section of 10×10 cm was cut from the uppermost peat of 14 sites using a sharp knife and multiple monoliths were collected from three sites ( Table 1). The cores were wrapped in plastic and stored in cool conditions until subsampling. A proximal sample of Hekla 1845 tephra was collected from a soil section (64.04 N°, 19.32 W°) in Iceland in 2018 for improving the proximal geochemistry dataset. This tephra was identified in the field based on its physical properties (colour and grain size) as well as its stratigraphic position below the Katla 1918 tephra marker layer. A sample covering the full thickness (1 cm) of the tephra layer was collected into a plastic bag and stored refrigerated until laboratory processing.
Additionally, the Hekla 1947 tephra sample collection of Salmi (1948) was reinvestigated in 2016. This collection consists of six tephra samples that had been recovered from snow cover and various other surfaces shortly after the fall-out and stored in sealed glass vials ever since (Salmi 1948). The collection sites of these samples are shown in Fig. 1.

Tephra extraction and processing
A subsample of one of the bottled Hekla 1947 samples (Västanfjärd) was treated with heavy liquid to determine the density range of the tephra shards. Sodium polytungstate solutions with densities of 2.2-2.7 g/cm 3 in 0.1 g/cm 3 steps were prepared and tephra shards retained in each density fraction (from < 2.2 g/cm 3 to > 2.7 g/cm 3 ) were mounted on glass slides with Canada Balsam and counted under a polarising microscope. Shards were also photographed for documentation, and their longest axis was measured to ascertain the average particle size of the tephra. All six samples in the collection were cleaned by floating off impurities using heavy liquid with a density of 2.2 g/cm 3 and subsequently mounted in epoxy resin for electron probe microanalysis (EPMA).
The proximal sample of Hekla 1845 tephra was cleaned by sieving with 120 and 63 µm meshes and the fraction remaining on the 63 µm sieve was mounted in an epoxy stub for EPMA.
The second step included a refined investigation of the peat samples based on the examination of the Hekla 1947 tephra collection of Salmi (1948). New 3 cm-long contiguous subsamples were extracted from each peat monolith and examined using laboratory methods that were adjusted to the results gained from the investigation of the Hekla 1947 tephra collection (see Results section). Sample volume was increased to ca. 10 cm 3 of fresh peat for every centimetre of the sampled depth range to increase the chances of finding even scarce shards. Samples were additionally sieved with a 25 µm mesh and heavy liquid densities were adjusted to extract the 2.3-2.6 g/cm 3 fraction.
Samples for EPMA were chosen from the depth of peak shard concentrations in the peat cores and treated with acid digestion instead of combustion to avoid geochemical alteration of volcanic glass at high temperatures (Pilcher and Hall, 1992;Dugmore et al., 1995). Tephra shards were then handpicked using a micromanipulator and mounted in epoxy resin.

EPMA
Major element composition of 6-27 tephra shards in each sample was determined by wavelength dispersive spectrometry on either JEOL JXA-8230 SuperProbe at the Institute of Earth Science (IES), University of Iceland or Cameca SX100 at the Tephra Analysis Unit (TAU) of the University of Edinburgh. Two samples (Punkaharju and KAN90) were analysed at both IES and TAU to confirm that the two instruments produce  (Table S1). Information on the respective electron microprobes and their configuration as well as analysis settings is available in supporting material (Table S1) and in Hayward (2012).

Radiocarbon dating
The Kivihypönneva site was selected for radiocarbon dating due to its high number of cryptotephra horizons. Moss (Sphagnum sp.) stems and leaves were picked for radiocarbon dating from the peat at 80-81 cm depth to determine a maximum age for the investigated sequence. Radiocarbon dating was performed at the Aarhus AMS Centre. The obtained radiocarbon age was calibrated using the online version of OxCal 4.3 (Bronk Ramsey, 2009) with the IntCal-13 calibration curve (Reimer et al., 2013).

Results
Hekla 1947 tephra collected by Salmi (1948) The Hekla 1947 tephra in Finland consists of colourless to brown shards with an average grain size of around 80 µm ( Fig. 3A), much larger than previously reported (Salmi, 1948). The density range of the glass grains is 2.2-2.7 g/cm 3 while the majority of them fall within the density range of 2.5-2.6 g/cm 3 (Fig. 3B). Colourless shards are present only in the 2.2-2.4 g/cm 3 density fraction, whereas brown shards exhibit a full range of densities ( Fig. 4A-D). The colour of each analysed shard in two samples (Punkaharju and Kuusjoki) was recorded during the EPMA sessions, but no correlation was found between the chemical composition and the shard colour (Table S1). In the Finnish samples, the SiO 2 content of the tephra ranges from 60 to 64.5 wt%, consistent with results from the proximal area (Tables 2  and S1). However, glass grains with SiO 2 > 69 wt% that have been recorded at proximal sites (Larsen et al., 1999), are seemingly absent from the Finnish samples.

Proximal sample of Hekla 1845 tephra
We analysed 15 shards of proximal Hekla 1845 tephra with an electron microprobe. The majority of the shards are of andesitic composition with an SiO 2 content of 57-62 wt%, whereas three shards are basaltic, and one shard is rhyolitic (Table S1). Tephra deposits from Hekla are known to have a range in chemical composition from basaltic to rhyolitic (Sigvaldason, 1974;Larsen and Thorarinsson, 1977;Guðmundsdóttir et al., 2011a;Jónsson et al., 2020).

Cryptotephra from peatland sites in Finland
Cryptotephra was identified in nine of our 25 research sites where it forms 19 stratigraphically discrete deposits that are composed of homogeneous tephra populations ( Fig. 5 and Tables 2 and S1). The stratigraphy as well as the tephra shard concentrations of the deposits are shown in Fig. 5. The highest number of cryptotephra horizons was found in Kivihypönneva, where four separate layers were detected at the depths of 34, 52, 58 and 67 cm ( Fig. 5 and Tables 2 and S1). Two sites, Pervarvikonneva and Kananiemensuo, contain three cryptotephra deposits each, at depths of 25, 60 and 65 cm and 40, 45 and 90 cm, respectively. Two cryptotephra deposits are present at three of the sites, Stormossen (at 24 and 58 cm depth), Rehtsuo (at 32 and 55 cm depth) and Suovanalanen (at 51 and 83 cm depth). At Haapasuo, Tarilampi and Hanhisuo, just one cryptotephra layer was detected at 57, 29 and 60 cm depths, respectively.
We were able to geochemically characterise 12 out of the 19 detected cryptotephra deposits and confirm the presence of four separate cryptotephra layers of Icelandic origin at our sites (Tables 2 and S1). The youngest geochemically identified cryptotephra forms a horizon of colourless rhyolitic glass with shard concentrations ranging from 15 to > 500 shards/cm 3 at six sites at depths of 29-60 cm (Fig. 5). This layer was found for the first time at two of these sites, Kivihypönneva and Kananiemensuo, whereas the occurrence and geochemical composition of this cryptotephra in Rehtsuo, Haapasuo, Tarilampi and Hanhisuo has already been published (Kalliokoski et al., 2019). EPMA of 17 individual shards indicates an Icelandic provenance for this tephra and plotting the results on a total alkali-silica (TAS) diagram reveals that the glass composition matches the geochemical envelope of the proximal products of the Askja volcano (Table 2 and Fig. 6A).
The second youngest geochemically identified cryptotephra horizon is present at just one site, Kananiemensuo, where it is positioned at the depth of 45 cm, approximately 5 cm below the Askja horizon (Fig. 5). This deposit consists of a small amount (<50 shards/cm 3 ) of light brown to dark brown tephra shards ( Fig. 4E-F). Major element characterisation of eight shards from this layer shows a geochemical affinity with the andesitic proximal products of the Hekla volcano when the results are plotted on a TAS diagram ( Fig. 6A and Table 2).
The second oldest geochemically characterised cryptotephra layer occurs in Kivihypönneva at 58 cm depth, in Pervarvikonneva at 60 cm depth and in Kananiemensuo at 90 cm depth. The peak shard concentration range for this layer is 60-80 shards/cm 3 and the shards are light brown with varying morphologies (Fig. 4G-H). EPMA results were obtained from 7-19 individual glass grains per layer and they plot within the geochemical envelope of andesitic-dacitic Hekla products on the TAS diagram (Tables 2 and S1, Fig. 6A).
The oldest geochemically fingerprinted cryptotephra layer is present at two sites: in Kivihypönneva at 67 cm depth and in Pervarvikonneva at 65 cm depth. It comprises a mixture of brown and colourless thin glass grains with fluted and vesicular morphologies (Figs 5 and 4I-J). Peak shard concentrations of this layer are 52 shards/cm 3 in Pervarvikonneva and 98 shards/cm 3 in Kivihypönneva. Despite the relatively high shard concentrations, EPMA succeeded for just 6 and 10 glass grains in Pervarvikonneva and Kivihypönneva, respectively, due to very thin shard walls ( Table 2). Inspection of the results on a TAS diagram shows that this layer represents dacitic products of the Hekla volcano (Fig. 6A).
Geochemical characterisation of seven other cryptotephra deposits at our sites failed, either due to a lack of material for the EPMA sample preparation procedure or loss of the scarce and tiny shards during the final steps of sanding and polishing the EPMA mounts. Based on the stratigraphic position of these deposits and characteristics of the tephra shards (Fig. 5), it is likely that they originate from the same Icelandic volcanoes as the geochemically identified layers in this study. However, geochemical fingerprinting is necessary for assigning them to specific eruptions.

Radiocarbon dating
Radiocarbon dating of a peat sample (lab. ID AAR-30738) from 80 to 81 cm depth at Kivihypönneva, ca. 13 cm below the oldest cryptotephra layer, gives a 14 C BP age of 1947 ± 24 and a calibrated age of 1-125 AD (2σ) for that depth (Fig. 5). This indicates that all the four identified cryptotephra layers in this study were deposited during the past two millennia.

Hekla 1947 tephra samples
The major element data of the Hekla 1947 tephra consist of 115 electron microprobe point-analyses of single shards (Table S1) and comprise a robust dataset for comparisons of future cryptotephra findings in the distal area. The analysed shards have not been subjected to possible geochemical alteration in an acidic environment, neither during preservation within the peat/ sediment matrix (Pollard et al., 2003) nor in laboratory procedures (Blockley et al., 2005;Cooper et al., 2019a). They therefore represent pristine volcanic glass of the Hekla 1947 eruption and provide a reference point for assessing the degree of geochemical alteration in tephra shards that are sourced from the same eruption and retrieved from environmental records.
The Hekla 1947 eruption (VEI 4) produced 0.18 km 3 of tephra in total (Thorarinsson, 1954). The explosive phase lasted for about 8 h and produced grey brownish tephra with SiO 2 values of 61-64 wt% during the Plinian opening stage (Thorarinsson, 1954;Thorsteinsdóttir et al., 2015). The tephra erupted in later stages was of a darker colour and contained 56-58 wt% of SiO 2 . The shards that were dispersed to Finland represent the Plinian phase based on their SiO 2 content of 60-64.5 wt%. The TiO 2 content of the Hekla 1947 tephra in Finland is constant at around 0.90 wt% (Fig. 7). Geochemical results from Iceland and Ireland show more scatter on bivariate plots, possibly indicating that the full range of eruption products was analysed there. When the Finnish results are compared with results from Iceland, Ireland and Britain, the highest compatibility is found between the glass compositions from Finland and Northern Ireland (Fig. 7).

Cryptotephra deposits in the Finnish peatlands
Geochemistry of the cryptotephra layers at Finnish sites was compared with the composition of proximal products of all the known Hekla and Askja eruptions of the past two millennia to determine their source eruptions (for a complete eruption list see Thorarinsson, 1967;Larsen et al., 1999Larsen et al., , 2020. For the sake of clarity, only the proximal tephras with compositions that most closely resemble the geochemistry of the Finnish deposits are shown in the bivariate plots of major element ratios to highlight the similarities and differences between these ( Fig. 6B-C).

Askja 1875 AD
The uppermost geochemically fingerprinted cryptotephra layer at our sites consists of rhyolitic glass originating from the Askja 1875 eruption based on its composition ( Fig. 6A-C, Table S1). The diagnostic feature of the Askja 1875 tephra is its high TiO 2 content of 0.7-1.0 wt% (Table S1), given that no other rhyolitic Icelandic tephras with TiO 2 > 0.7 wt% are known from the historical period (e.g. Larsen et al., 1999). Furthermore, the 1875 eruption is the only recorded historical Askja event that produced silicic tephra, and the common presence of this tephra in Norway, Sweden and Finland (Wastegård, 2005;Davies et al., 2007;Carey et al., 2010;Kalliokoski et al., 2019) supports the correlation made here.

Hekla 1845 AD
A cryptotephra layer in Kananiemensuo at the depth of 45 cm is correlated to the 1845 eruption of Hekla based on both its geochemistry and its stratigraphic position < 5 cm below the Askja 1875 tephra (Fig. 5). Comparison with the new proximal data of the Hekla 1845 tephra reveals a strong similarity between the deposits in Finland, Northern Ireland (Watson et al., 2015) and Iceland (Table 2 and Fig. 6B-C). In Finland this tephra is andesitic with an SiO 2 content of 60-62 wt%, while at the proximal site in Iceland basaltic glass and one rhyolitic shard were also analysed (Table S1), and in Northern Ireland a minor rhyolitic component has been identified alongside the predominant andesitic component (Watson et al., 2015). The Hekla 1845 tephra can be clearly distinguished from other known historical products of Hekla by its FeO/K 2 O ratio (Fig. 6B).

Hekla 1510 AD
Andesitic-dacitic cryptotephra of Hekla origin was identified at our southernmost site in Kananiemensuo as well as in west-central Finland in Kivihypönneva and Pervarvikonneva. Both the geochemical composition and stratigraphic position of this layer below the Askja 1875 and Hekla 1845 deposits support a correlation to the Hekla 1510 eruption (Fig. 5 and Tables 2 and S1). The geochemical compositions of the Hekla 1947 and Hekla 1510 tephras are almost identical based on the proximal data (Larsen et al., 1999), but the stratigraphic position of this layer at Finnish sites below the Askja 1875 cryptotephra horizon excludes the 1947 eruption as its source. The lower FeO content (< 8 wt%) of the Hekla 1510 tephra distinguishes it from historical andesitic Hekla products  (Fig. 6B-C).

Hekla 1158 AD
The oldest identified cryptotephra layer at our sites is the dacitic Hekla 1158 layer in Kivihypönneva and Pervarvikonneva, where it is positioned <10 cm below the Hekla 1510 cryptotephra horizon. The Hekla 1158 tephra can be distinguished from the other historical silicic Hekla tephra, the Hekla 1104, based on its significantly lower (66.4-68.2 wt%) SiO 2 and higher FeO content ( Fig. 6B-C; Larsen et al., 1999). Correlation of this cryptotephra layer in Finland to the Hekla 1158 eruption is also supported by its earlier identification in northern parts of Norway and Sweden (Pilcher et al., 2005;Balascio et al., 2011;Watson et al., 2016;Cooper et al., 2019b).

Absence of the Hekla 1947 tephra from Finnish sites
We searched for the Hekla 1947 tephra by a thorough and repeated investigation of peat samples. We modified the routine laboratory methods based on investigation of Hekla 1947 shard properties but failed to locate any tephra shards that could be confidently assigned to the 1947 eruption. At two sites in west-central Finland, in the vicinity of an unconfirmed contemporary account of Hekla 1947 tephra fall-out, a trace amount (<15 shards/cm 3 ) of light brown to brown cryptotephra shards (Fig. 4K-L) was detected in the uppermost peat, in Pervarvikonneva at 27 cm and in Kivihypönneva at 34 cm depth (sites 10 and 12 in Figs 1 and 5). At Kivihypönneva this tephra occurs ca. 15 cm above the Askja 1875 tephra. Unfortunately, the geochemical characterisation of the shards failed due to a lack of material, and correlations to source eruptions could thus not be established. Absence of the Hekla 1947 tephra from Finnish sites is a surprising result, as its fall-out in Finland has long been considered a classic example of the potential for a smaller scale eruption to form a tephra isochrone in the far-distal field. Given that this research resulted in the identification of cryptotephra deposits from four older eruptions, we can rule out the possibility that our cores would have been too short to record the Hekla 1947 fall-out or that the chosen methodology would have impeded detection of volcanic glass grains. The absence of the Hekla 1947 tephra from our sites could be explained by a patchy fall-out or post-depositional horizontal movement of shards that were deposited on snow cover. It was not possible to verify a connection between precipitation and the Hekla 1947 tephra fall-out in this study, but inspection of precipitation maps with locations of eyewitness reports of tephra fall-out reveals that tephra fell mainly south of the zone of highest precipitation and also in areas that experienced no or minimal (< 1 mm/24 h) precipitation ( Fig. 2A-C). The snow-depth map from 30 March 1947 reveals that snow was present almost throughout the fall-out zone (Fig. 2D). It has been suggested that reworking processes by meltwater and winds concentrate tephra shards in microtopographic hollows over the melting snowpack surface, rendering the layer patchy on a small scale (Bergman et al., 2004). In the case of the Hekla 1947 tephra, this is also supported by observations of the visible fall-out in Finland commonly occurring in depressions in the snowpack (Salmi, 1948). Additionally, post-depositional horizontal movement of tephra shards over peat surface has been experimentally verified (Payne and Gehrels, 2010). On the other hand, we have carefully investigated nearly 50 surface cores from over 30 peatland and lake sites within the fall-out zone without finding the Hekla 1947 tephra. Even if post-depositional horizontal redistribution of cryptotephra could explain the lack of Hekla 1947 deposits in some of the coring sites, the same processes should have concentrated the tephra into others and at least some of the 50 cores would be expected to contain a detectable layer.

Preservation potential of volcanic glass in acidic environments
Poor post-depositional preservation of cryptotephra is another possible explanation for the absence of the Hekla 1947 tephra from Finnish sites. It has been suggested that geochemical alteration or even a total dissolution of small and vesicular volcanic glass shards, and basaltic glass specifically, takes place in acidic environments, such as peat bogs, or during acid digestion laboratory procedures (Pollard et al., 2003;Blockley et al., 2005). In Finland the environment is prone to acidification due to the prevailing base-poor and resistant granitic bedrock. Peat pH has been measured to range from 2.8-4.4 in raised peat bogs to 4.0-5.2 in fens (Martikainen et al., 1993;Laine et al., 1995;Jaatinen et al., 2005). Considering this, poor preservation of andesitic-dacitic shards in an acidic bog environment might contribute to the apparent absence of the Hekla 1947 tephra in Finland. The degree of susceptibility of volcanic glass to geochemical alteration in acidic and basic conditions is still a matter of debate and two recently conducted experiments on the effects of strong acids and bases on volcanic glass particles during laboratory procedures give contradictory results (Monteath et al., 2019;Cooper et al., 2019a). The results of Monteath et al. (2019) do not support the idea of readily soluble volcanic glass, whereas Cooper et al. (2019a) found the geochemical alteration in andesitic and basaltic glass to be significant enough to result in possible miscorrelations when tracing the source eruptions of cryptotephra. According to Cooper et al. (2019a), a further implication of the observed alteration of low-silica glass would be its poorer stability in acidic peat bogs over longer timescales, which might partly explain the scarcity of basaltic glass in North European cryptotephra records (e.g. Wastegård and Davies 2009;Lawson et al., 2012). However, in the case of the andesitic-dacitic Hekla 1947 tephra, its residence time in an acidic peat environment would be short; merely 67 years at the time of coring. A total dissolution of the shards is therefore considered unlikely, especially when the older andesitic Hekla 1845 and andesitic-dacitic Hekla 1510 tephras, Figure 6. (A) Composition of Finnish cryptotephra deposits and geochemical envelopes of Icelandic volcanic systems shown on a total alkali-silica diagram. Nomenclature for volcanic rocks from Le Bas et al. (1986). High-and low-alkali products are separated by the Kuno line (Kuno, 1966) drawn in black. Data for the Icelandic volcanic systems are from Larsen et al., 1999, andadditionally from Sigvaldason, 1979;Prestvik, 1985;Steinthorsson et al., 1985;Larsen et al., 2001;Larsen et al., 2002;Eiríksson et al., 2004;Sverrisdóttir, 2007;Guðmundsdóttir et al., 2011b;Óladóttir et al., 2011;Kalliokoski et al., (Table S1). [Color figure can be viewed at wileyonlinelibrary.com] that have been calculated to have approximately the same theoretical stability as Hekla 1947 (Pollard et al., 2003), are preserved at the Finnish sites. It is, however, possible that the preservation potential of tephra at any single site may vary through time due to changing environmental conditions. For example, it has been shown that air pollution inducing acidification of surface waters in Fennoscandia commenced in the 1940s and accelerated in the 1960s with increased combustion of fossil fuels (Tolonen and Jaakkola, 1983;Renberg, 1990). At that time, the Hekla 1947 tephra would still have resided at a shallow depth in the peat where acid rain likely resulted in elevated acidity, contributing to higher leaching rates of volcanic glass. The draining of peatlands has also contributed to the increased acidity of near-surface peat through lowering of the water table (Laine et al., 1995;Jaatinen et al., 2005). As the older tephra deposits, such as Hekla 1845 and 1510, lie deeper in the peat stratigraphy they would have been less affected by the acidification trend of the previous decades, which could explain their better preservation.
When the geochemical composition of the pristine Hekla 1947 shards in this study is compared with the composition of the Hekla 1947 cryptotephra layer in Northern Ireland (Rea et al., 2012), no sign of geochemical alteration is detected in the Northern Irish shards that were retrieved from peat bogs and extracted using acid digestion. This is in line with the observations of Watson et al. (2015) who found no significant geochemical alteration in any of the rhyolitic Hekla 1104 or trachydacitic Sn-1 cryptotephra shards from lake (pH 7) or bog sites (pH 5.9) in Sweden. It could be argued that surface waters and peat bogs in Ireland and Britain would be generally less acidic than their Finnish and Swedish counterparts due to more base-rich bedrock, and thus favour the preservation of volcanic glass. However, atmospheric deposition of SO 2 emissions has driven acidification of surface waters in Ireland as well (Leira et al., 2007) and peat pH of 3.5-8.5 has been measured in peatlands in Ireland and the UK (Wheeler and Proctor, 2000). Therefore, the acidic conditions of the Finnish peatlands alone do not sufficiently well explain the absence of the Hekla 1947 at Finnish sites. Whatever the reason(s) for the absence of Hekla 1947 tephra from the Finnish sites may be, our results imply that the chances of finding Hekla 1947 tephra in Finland are poor and it is unlikely to become an important isochrone in the region.
Dispersal patterns of Hekla 1158, Hekla 1510, Hekla 1845 and Hekla 1947 tephras The Hekla 1158 eruption (VEI 4) began on 19 January and produced 0.2-0.6 km 3 of tephra (Janebo et al., 2016). In Iceland, the dispersal axis of the 1158 tephra extends northeast from Hekla (Larsen et al., 1999), and earlier identifications of this tephra in northern Norway (Pilcher et al., 2005;Balascio et al., 2011) and Sweden (Watson et al., 2016;Cooper et al., 2019b) as well as its absence from central and southern Sweden suggest that it reached Finland via a northerly transport route (Fig. 8). Hekla 1158 tephra was identified at just two peatland sites in this study. However, its dispersal area in Finland may well be wider than reported here, given that it was found in two of our three northernmost sites and no peatlands further north were investigated.
The Hekla 1845 eruption (VEI 4) started on 2 September and produced a minimum of 0.13 km 3 of tephra during its 1 h-long explosive opening phase (Guðnason et al., 2018). In Iceland the axis of tephra fall extends east-southeast of Hekla (Guðnason et al., 2018). Based on eyewitness reports from the Shetland Islands, Orkney and the Faroe Islands (Thorarinsson, 1981;Guðnason et al., 2018) as well as earlier identification of this tephra in the Faroe Islands (Wastegård, 2002), Northern Ireland (Watson et al., 2015) and Wales (Watson et al., 2017), we postulate the transport pathway of this tephra to be similar to that of the Hekla 1947 tephra (Fig. 8). Dispersal of the Hekla 1845 tephra further east towards southern Finland along a curving pathway is also supported by observations of a volcanic haze in Sassnitz on the German coast of the Baltic Sea 4-5 days after the eruption ( Fig. 8; Boll 1846 in Thorarinsson, 1981).
The Hekla 1510 eruption (VEI 4) began on 25 July and produced around 0.33 km 3 of tephra (Thorarinsson, 1967). Occurrence of the Hekla 1510 tephra has been reported to be patchy within its dispersal area in the UK (Dugmore et al., 1995;Lawson et al., 2012). In Finland, significantly further away from the source volcano, the occurrence of this tephra would be expected to be even more sporadic. Interestingly, our results indicate that Hekla 1510 tephra is quite widespread in Finland, and some of the unidentified deposits at our sites may well derive from the same eruption based on their stratigraphical position and the appearance of shards. For example, just one analysis was obtained from a cryptotephra deposit in Rehtsuo at 55 cm depth (Fig. 5), but the geochemical composition matches that of the Hekla 1510 tephra (Table S1).
The presence of Hekla 1845 and Hekla 1510 tephras in Ireland, the UK and Finland and their apparent absence from the rest of Fennoscandia and the European mainland indicates that their dispersal patterns are probably as complex as that of the Hekla 1947 tephra (Figs 1 and 8). However, using the absence of a cryptotephra to determine its dispersal in the distal area is not straightforward due to the many post-depositional processes and preservation issues that may impede cryptotephra detection (see Watson et al., 2015 and references therein). This holds true especially for tephra originating from smaller scale eruptions, since tephra deposits consisting of scarce shards may be masked by the nearby presence of other more prominent layers. For example, cryptotephra originating from the Askja 1875 eruption (VEI 5) that produced 1.8 km 3 of tephra (Carey et al., 2010), forms common deposits with high shard concentrations throughout Sweden, Norway and Finland (Wastegård, 2005;Davies et al., 2007;Carey et al., 2010;Kalliokoski et al., 2019). If shard concentration profiles indicate the presence of a diffuse tephra layer and a sample for EPMA comprises only the depth of maximum shard concentration, tephra from smaller eruptions that are relatively close in age to the dominant eruption may be interpreted as a result of contamination or post-depositional reworking of the primary fall-out deposit and therefore remain unanalysed. As the technique of concentrating cryptotephra shards for EPMA by using a micromanipulator is relatively new, obtaining robust geochemical results from cryptotephra deposits with scarce shards has previously been very difficult and earlier research has focused mainly on the layers with the highest shard concentrations.

Significance for Finnish tephrochronology
The presence of four historical Icelandic cryptotephra layers in Finnish peat bogs indicates a great potential for using tephrochronology as a dating method in environmental research in the region. The most common cryptotephra layer in southern and middle Finland, the Askja 1875 tephra, was geochemically identified at six sites and was possibly present in two others. Hekla 1510 and Hekla 1158 tephras were found at three and two sites, respectively. However, their occurrence in Finland may be more common than reported here. Some of the detected cryptotephra horizons remain as yet unanalysed, and at more than half of the research sites only the uppermost 50 cm of peat were investigated, whereas the Hekla 1510 and Hekla 1158 cryptotephras were found deeper than 50 cm. In contrast, the Hekla 1845 cryptotephra was identified only in the southernmost site and, considering its occurrence depth just a couple of centimetres below the Askja 1875 horizon, it would probably have been detected had it been present at the other investigated sites. We suggest that at least the Askja 1875, Hekla 1510 and Hekla 1158 tephras may all become important marker layers in environmental research focusing on the past millennium, an era of increasing anthropogenic influence in Finland.
In this study, the Hekla 1845 and Hekla 1510 cryptotephras are identified for the first time outside of Ireland, the UK and the Faroes and their confirmed dispersal is extended significantly further east. The occurrence of Hekla 1845, Hekla 1510 and Hekla 1158 horizons in Finland is also an important implication for the potential of Icelandic eruptions of VEI ≤ 4 to create widely dispersed isochrones in the distal area, which allow inter-regional correlations to be made. Identification of four historical Icelandic tephras in Finland, as well as the fallout of Hekla 1947 tephra (Salmi, 1948), the detection of the Eyjafjallajökull 2010 tephra in Finnish airspace (Davies et al., 2010) and the identification of the Grímsvötn 2011 tephra in surface layer air in southern Finland (Kerminen et al., 2011) highlight that volcanic ash from Icelandic eruptions is frequently carried to Finland. Furthermore, our results demonstrate that in addition to a direct eastward dispersal from Iceland, ash clouds may travel to Finland along both northerly and southerly transport routes. This implies that assuming a direct eastward transport route for Icelandic volcanic ash and estimating the frequency of volcanic ash events in Finnish airspace by extrapolating from the wellestablished Swedish and Norwegian tephrochronologies would give misleading results. This finding stresses the importance of further tephrochronological research in Finland and supports the predictions from simulations of volcanic ash transport of a high probability (46%) of tephra fall-out at Helsinki airport even after a moderate Hekla eruption (Leadbetter and Hort, 2011).

Conclusions
Hekla 1947 tephra was searched from 25 peatlands in southern and central Finland. Despite careful inspection of 47 surface peat cores and monoliths, no cryptotephra deposits could be attributed to the Hekla 1947 eruption. On the other hand, Hekla 1845, Hekla 1510 and Hekla 1158 cryptotephras were identified for the first time in Finland and their known dispersal areas were extended significantly eastwards. The presence of these tephras in Finland demonstrates that isochrones from Icelandic smaller scale eruptions are more extensive than previously realised. In addition, Askja 1875 cryptotephra was identified at two new sites in Finland. This result highlights the importance of Askja 1875 as the most prominent historical cryptotephra horizon in the region. Our results also reveal that in addition to a direct eastward dispersal route the ash clouds from Icelandic volcanic eruptions can be Hekla 1158 tephras in Northern Europe. Circles mark tentative correlations and dots geochemically confirmed cryptotephra deposits or contemporary eyewitness reports of fall-out. Axes of tephra fall within Iceland are from Larsen et al., 1999 andGuðnason et al., 2018. The arrow marking the main dispersal axis of Hekla 1947 tephra based on prevailing wind directions is from Salmi (1948). Data for sites of cryptotephra detection and eyewitness reports are from Dugmore et al., 1995;Pilcher et al., 1996;Wastegård, 2002;Pilcher et al., 2005;Balascio et al., 2011;Reilly and Mitchell, 2015;Watson et al., 2015Watson et al., , 2016Watson et al., and 2017Guðnason et al., 2018;Cooper et al., 2019b. [Color figure can be viewed at wileyonlinelibrary.com] transported to Finland along complex southerly and northerly dispersal routes. Absence of the Hekla 1947 tephra from its previously inferred fall-out zone in Finland remains unexplained and could be an indication of either post-depositional reworking of shards or poorer preservation potential of andesitic glass in peatlands during the 20th century due to increasingly acidic conditions caused by air pollution-induced acid rain and lowering of the water table. The identification of four historical Icelandic cryptotephras in Finnish peatlands demonstrates the potential of cryptotephra studies in Finland and is an important step towards building a Finnish tephrochronological framework.

Supporting information
Additional supporting information may be found in the online version of this article at the publisher's web-site.