diff --git a/pyerrors/input/dobs.py b/pyerrors/input/dobs.py
new file mode 100644
index 00000000..498e88d8
--- /dev/null
+++ b/pyerrors/input/dobs.py
@@ -0,0 +1,873 @@
+from collections import defaultdict
+import gzip
+import lxml.etree as et
+import getpass
+import socket
+import datetime
+import json
+import warnings
+import numpy as np
+from ..obs import Obs
+from ..obs import _merge_idx
+from ..covobs import Covobs
+from .. import version as pyerrorsversion
+
+
+# Based on https://stackoverflow.com/a/10076823
+def _etree_to_dict(t):
+    """ Convert the content of an XML file to a python dict"""
+    d = {t.tag: {} if t.attrib else None}
+    children = list(t)
+    if children:
+        dd = defaultdict(list)
+        for dc in map(_etree_to_dict, children):
+            for k, v in dc.items():
+                dd[k].append(v)
+        d = {t.tag: {k: v[0] if len(v) == 1 else v
+                     for k, v in dd.items()}}
+    if t.attrib:
+        d[t.tag].update(('@' + k, v)
+                        for k, v in t.attrib.items())
+    if t.text:
+        text = t.text.strip()
+        if children or t.attrib:
+            if text:
+                d[t.tag]['#data'] = [text]
+        else:
+            d[t.tag] = text
+    return d
+
+
+def _dict_to_xmlstring(d):
+    if isinstance(d, dict):
+        iters = ''
+        for k in d:
+            if k.startswith('#'):
+                for la in d[k]:
+                    iters += la
+                iters = '<array>\n' + iters + '<%sarray>\n' % ('/')
+                return iters
+            if isinstance(d[k], dict):
+                iters += '<%s>\n' % (k) + _dict_to_xmlstring(d[k]) + '<%s%s>\n' % ('/', k)
+            elif isinstance(d[k], str):
+                if len(d[k]) > 100:
+                    iters += '<%s>\n ' % (k) + d[k] + ' \n<%s%s>\n' % ('/', k)
+                else:
+                    iters += '<%s> ' % (k) + d[k] + ' <%s%s>\n' % ('/', k)
+            elif isinstance(d[k], list):
+                for i in range(len(d[k])):
+                    iters += _dict_to_xmlstring(d[k][i])
+            elif not d[k]:
+                return '\n'
+            else:
+                raise Exception('Type', type(d[k]), 'not supported in export!')
+    else:
+        raise Exception('Type', type(d), 'not supported in export!')
+    return iters
+
+
+def _dict_to_xmlstring_spaces(d, space='  '):
+    s = _dict_to_xmlstring(d)
+    o = ''
+    c = 0
+    cm = False
+    for li in s.split('\n'):
+        if li.startswith('<%s' % ('/')):
+            c -= 1
+            cm = True
+        for i in range(c):
+            o += space
+        o += li + '\n'
+        if li.startswith('<') and not cm:
+            if not '<%s' % ('/') in li:
+                c += 1
+        cm = False
+    return o
+
+
+def create_pobs_string(obsl, name, spec='', origin='', symbol=[], enstag=None):
+    """Export a list of Obs or structures containing Obs to an xml string
+    according to the Zeuthen pobs format.
+
+    Tags are not written or recovered automatically. The separator | is removed from the replica names.
+
+    Parameters
+    ----------
+    obsl : list
+        List of Obs that will be exported.
+        The Obs inside a structure have to be defined on the same ensemble.
+    name : str
+        The name of the observable.
+    spec : str
+        Optional string that describes the contents of the file.
+    origin : str
+        Specify where the data has its origin.
+    symbol : list
+        A list of symbols that describe the observables to be written. May be empty.
+    enstag : str
+        Enstag that is written to pobs. If None, the ensemble name is used.
+    """
+
+    od = {}
+    ename = obsl[0].e_names[0]
+    names = list(obsl[0].deltas.keys())
+    nr = len(names)
+    onames = [name.replace('|', '') for name in names]
+    for o in obsl:
+        if len(o.e_names) != 1:
+            raise Exception('You try to export dobs to obs!')
+        if o.e_names[0] != ename:
+            raise Exception('You try to export dobs to obs!')
+        if len(o.deltas.keys()) != nr:
+            raise Exception('Incompatible obses in list')
+    od['observables'] = {}
+    od['observables']['schema'] = {'name': 'lattobs', 'version': '1.0'}
+    od['observables']['origin'] = {
+        'who': getpass.getuser(),
+        'date': str(datetime.datetime.now())[:-7],
+        'host': socket.gethostname(),
+        'tool': {'name': 'pyerrors', 'version': pyerrorsversion.__version__}}
+    od['observables']['pobs'] = {}
+    pd = od['observables']['pobs']
+    pd['spec'] = spec
+    pd['origin'] = origin
+    pd['name'] = name
+    if enstag:
+        if not isinstance(enstag, str):
+            raise Exception('enstag has to be a string!')
+        pd['enstag'] = enstag
+    else:
+        pd['enstag'] = ename
+    pd['nr'] = '%d' % (nr)
+    pd['array'] = []
+    osymbol = 'cfg'
+    if not isinstance(symbol, list):
+        raise Exception('Symbol has to be a list!')
+    if not (len(symbol) == 0 or len(symbol) == len(obsl)):
+        raise Exception('Symbol has to be a list of lenght 0 or %d!' % (len(obsl)))
+    for s in symbol:
+        osymbol += ' %s' % s
+    for r in range(nr):
+        ad = {}
+        ad['id'] = onames[r]
+        Nconf = len(obsl[0].deltas[names[r]])
+        layout = '%d i f%d' % (Nconf, len(obsl))
+        ad['layout'] = layout
+        ad['symbol'] = osymbol
+        data = ''
+        for c in range(Nconf):
+            data += '%d ' % obsl[0].idl[names[r]][c]
+            for o in obsl:
+                num = o.deltas[names[r]][c] + o.r_values[names[r]]
+                if num == 0:
+                    data += '0 '
+                else:
+                    data += '%1.16e ' % (num)
+            data += '\n'
+        ad['#data'] = data
+        pd['array'].append(ad)
+
+    rs = '<?xml version="1.0" encoding="utf-8"?>\n' + _dict_to_xmlstring_spaces(od)
+    return rs
+
+
+def write_pobs(obsl, fname, name, spec='', origin='', symbol=[], enstag=None, gz=True):
+    """Export a list of Obs or structures containing Obs to a .xml.gz file
+    according to the Zeuthen pobs format.
+
+    Tags are not written or recovered automatically. The separator | is removed from the replica names.
+
+    Parameters
+    ----------
+    obsl : list
+        List of Obs that will be exported.
+        The Obs inside a structure have to be defined on the same ensemble.
+    fname : str
+        Filename of the output file.
+    name : str
+        The name of the observable.
+    spec : str
+        Optional string that describes the contents of the file.
+    origin : str
+        Specify where the data has its origin.
+    symbol : list
+        A list of symbols that describe the observables to be written. May be empty.
+    enstag : str
+        Enstag that is written to pobs. If None, the ensemble name is used.
+    gz : bool
+        If True, the output is a gzipped xml. If False, the output is an xml file.
+    """
+    pobsstring = create_pobs_string(obsl, name, spec, origin, symbol, enstag)
+
+    if not fname.endswith('.xml') and not fname.endswith('.gz'):
+        fname += '.xml'
+
+    if gz:
+        if not fname.endswith('.gz'):
+            fname += '.gz'
+
+        fp = gzip.open(fname, 'wb')
+        fp.write(pobsstring.encode('utf-8'))
+    else:
+        fp = open(fname, 'w', encoding='utf-8')
+        fp.write(pobsstring)
+    fp.close()
+
+
+def _import_data(string):
+    return json.loads("[" + ",".join(string.replace(' +', ' ').split()) + "]")
+
+
+def _check(condition):
+    if not condition:
+        raise Exception("XML file format not supported")
+
+
+class _NoTagInDataError(Exception):
+    """Raised when tag is not in data"""
+    def __init__(self, tag):
+        self.tag = tag
+        super().__init__('Tag %s not in data!' % (self.tag))
+
+
+def _find_tag(dat, tag):
+    for i in range(len(dat)):
+        if dat[i].tag == tag:
+            return i
+    raise _NoTagInDataError(tag)
+
+
+def _import_array(arr):
+    name = arr[_find_tag(arr, 'id')].text.strip()
+    index = _find_tag(arr, 'layout')
+    try:
+        sindex = _find_tag(arr, 'symbol')
+    except _NoTagInDataError:
+        sindex = 0
+    if sindex > index:
+        tmp = _import_data(arr[sindex].tail)
+    else:
+        tmp = _import_data(arr[index].tail)
+
+    li = arr[index].text.strip()
+    m = li.split()
+    if m[1] == "i" and m[2][0] == "f":
+        nc = int(m[0])
+        na = int(m[2].lstrip('f'))
+        _dat = []
+        mask = []
+        for a in range(na):
+            mask += [a]
+            _dat += [np.array(tmp[1 + a:: na + 1])]
+        _check(len(tmp[0:: na + 1]) == nc)
+        return [name, tmp[0:: na + 1], mask, _dat]
+    elif m[1][0] == 'f' and len(m) < 3:
+        sh = (int(m[0]), int(m[1].lstrip('f')))
+        return np.reshape(tmp, sh)
+    elif any(['f' in s for s in m]):
+        for si in range(len(m)):
+            if m[si] == 'f':
+                break
+        sh = [int(m[i]) for i in range(si)]
+        return np.reshape(tmp, sh)
+    else:
+        print(name, m)
+        _check(False)
+
+
+def _import_rdata(rd):
+    name, idx, mask, deltas = _import_array(rd)
+    return deltas, name, idx
+
+
+def _import_cdata(cd):
+    _check(cd[0].tag == "id")
+    _check(cd[1][0].text.strip() == "cov")
+    cov = _import_array(cd[1])
+    grad = _import_array(cd[2])
+    return cd[0].text.strip(), cov, grad
+
+
+def read_pobs(fname, full_output=False, gz=True, separator_insertion=None):
+    """Import a list of Obs from an xml.gz file in the Zeuthen pobs format.
+
+    Tags are not written or recovered automatically.
+
+    Parameters
+    ----------
+    fname : str
+        Filename of the input file.
+    full_output : bool
+        If True, a dict containing auxiliary information and the data is returned.
+        If False, only the data is returned as list.
+    separatior_insertion: str or int
+        str: replace all occurences of "separator_insertion" within the replica names
+        by "|%s" % (separator_insertion) when constructing the names of the replica.
+        int: Insert the separator "|" at the position given by separator_insertion.
+        None (default): Replica names remain unchanged.
+    """
+
+    if not fname.endswith('.xml') and not fname.endswith('.gz'):
+        fname += '.xml'
+    if gz:
+        if not fname.endswith('.gz'):
+            fname += '.gz'
+        with gzip.open(fname, 'r') as fin:
+            content = fin.read()
+    else:
+        if fname.endswith('.gz'):
+            warnings.warn("Trying to read from %s without unzipping!" % fname, UserWarning)
+        with open(fname, 'r') as fin:
+            content = fin.read()
+
+    # parse xml file content
+    root = et.fromstring(content)
+
+    _check(root[2].tag == 'pobs')
+    pobs = root[2]
+
+    version = root[0][1].text.strip()
+
+    _check(root[1].tag == 'origin')
+    file_origin = _etree_to_dict(root[1])['origin']
+
+    deltas = []
+    names = []
+    idl = []
+    for i in range(5, len(pobs)):
+        delta, name, idx = _import_rdata(pobs[i])
+        deltas.append(delta)
+        if separator_insertion is None:
+            pass
+        elif isinstance(separator_insertion, int):
+            name = name[:separator_insertion] + '|' + name[separator_insertion:]
+        elif isinstance(separator_insertion, str):
+            name = name.replace(separator_insertion, "|%s" % (separator_insertion))
+        else:
+            raise Exception("separator_insertion has to be string or int, is ", type(separator_insertion))
+        names.append(name)
+        idl.append(idx)
+    res = [Obs([d[i] for d in deltas], names, idl=idl) for i in range(len(deltas[0]))]
+
+    descriptiond = {}
+    for i in range(4):
+        descriptiond[pobs[i].tag] = pobs[i].text.strip()
+
+    _check(pobs[4].tag == "nr")
+
+    _check(pobs[5].tag == 'array')
+    if pobs[5][1].tag == 'symbol':
+        symbol = pobs[5][1].text.strip()
+        descriptiond['symbol'] = symbol
+
+    if full_output:
+        retd = {}
+        tool = file_origin.get('tool', None)
+        if tool:
+            program = tool['name'] + ' ' + tool['version']
+        else:
+            program = ''
+        retd['program'] = program
+        retd['version'] = version
+        retd['who'] = file_origin['who']
+        retd['date'] = file_origin['date']
+        retd['host'] = file_origin['host']
+        retd['description'] = descriptiond
+        retd['obsdata'] = res
+        return retd
+    else:
+        return res
+
+
+# Reading (and writing) dobs is not yet working properly:
+# we have to loop over root[2:] because each entry is a dobs
+# But maybe this is just a problem with Ben's implementation
+
+# this is based on Mattia Bruno's implementation at https://github.com/mbruno46/pyobs/blob/master/pyobs/IO/xml.py
+def import_dobs_string(content, noempty=False, full_output=False, separator_insertion=True):
+    """Import a list of Obs from a string in the Zeuthen dobs format.
+
+    Tags are not written or recovered automatically.
+
+    Parameters
+    ----------
+    content : str
+        XML string containing the data
+    noemtpy : bool
+        If True, ensembles with no contribution to the Obs are not included.
+        If False, ensembles are included as written in the file, possibly with vanishing entries.
+    full_output : bool
+        If True, a dict containing auxiliary information and the data is returned.
+        If False, only the data is returned as list.
+    separatior_insertion: str, int or bool
+        str: replace all occurences of "separator_insertion" within the replica names
+        by "|%s" % (separator_insertion) when constructing the names of the replica.
+        int: Insert the separator "|" at the position given by separator_insertion.
+        True (default): separator "|" is inserted after len(ensname), assuming that the
+        ensemble name is a prefix to the replica name.
+        None or False: No separator is inserted.
+    """
+
+    root = et.fromstring(content)
+
+    _check(root.tag == 'OBSERVABLES')
+    _check(root[0].tag == 'SCHEMA')
+    version = root[0][1].text.strip()
+
+    _check(root[1].tag == 'origin')
+    file_origin = _etree_to_dict(root[1])['origin']
+
+    _check(root[2].tag == 'dobs')
+
+    dobs = root[2]
+
+    descriptiond = {}
+    for i in range(3):
+        descriptiond[dobs[i].tag] = dobs[i].text.strip()
+
+    _check(dobs[3].tag == 'array')
+
+    symbol = []
+    if dobs[3][1].tag == 'symbol':
+        symbol = dobs[3][1].text.strip()
+        descriptiond['symbol'] = symbol
+    mean = _import_array(dobs[3])[0]
+
+    _check(dobs[4].tag == "ne")
+    ne = int(dobs[4].text.strip())
+    _check(dobs[5].tag == "nc")
+    nc = int(dobs[5].text.strip())
+
+    idld = {}
+    deltad = {}
+    covd = {}
+    gradd = {}
+    names = []
+    e_names = []
+    enstags = {}
+    for k in range(6, len(list(dobs))):
+        if dobs[k].tag == "edata":
+            _check(dobs[k][0].tag == "enstag")
+            ename = dobs[k][0].text.strip()
+            e_names.append(ename)
+            _check(dobs[k][1].tag == "nr")
+            R = int(dobs[k][1].text.strip())
+            for i in range(2, 2 + R):
+                deltas, rname, idx = _import_rdata(dobs[k][i])
+                if separator_insertion is None or False:
+                    pass
+                elif separator_insertion is True:
+                    if rname.startswith(ename):
+                        rname = rname[:len(ename)] + '|' + rname[len(ename):]
+                elif isinstance(separator_insertion, int):
+                    rname = rname[:separator_insertion] + '|' + rname[separator_insertion:]
+                elif isinstance(separator_insertion, str):
+                    rname = rname.replace(separator_insertion, "|%s" % (separator_insertion))
+                else:
+                    raise Exception("separator_insertion has to be string or int, is ", type(separator_insertion))
+                if '|' in rname:
+                    new_ename = rname[:rname.index('|')]
+                else:
+                    new_ename = ename
+                enstags[new_ename] = ename
+                idld[rname] = idx
+                deltad[rname] = deltas
+                names.append(rname)
+        elif dobs[k].tag == "cdata":
+            cname, cov, grad = _import_cdata(dobs[k])
+            covd[cname] = cov
+            if grad.shape[1] == 1:
+                gradd[cname] = [grad for i in range(len(mean))]
+            else:
+                gradd[cname] = grad.T
+        else:
+            _check(False)
+    names = list(set(names))
+
+    for name in names:
+        for i in range(len(deltad[name])):
+            deltad[name][i] = np.array(deltad[name][i]) + mean[i]
+
+    res = []
+    for i in range(len(mean)):
+        deltas = []
+        idl = []
+        obs_names = []
+        for name in names:
+            h = np.unique(deltad[name][i])
+            if len(h) == 1 and np.all(h == mean[i]) and noempty:
+                continue
+            deltas.append(deltad[name][i])
+            obs_names.append(name)
+            idl.append(idld[name])
+        res.append(Obs(deltas, obs_names, idl=idl))
+    print(mean, 'vs', res)
+    _check(len(e_names) == ne)
+
+    cnames = list(covd.keys())
+    for i in range(len(res)):
+        new_covobs = {name: Covobs(0, covd[name], name, grad=gradd[name][i]) for name in cnames}
+        if noempty:
+            for name in cnames:
+                if np.all(new_covobs[name].grad == 0):
+                    del new_covobs[name]
+            cnames_loc = list(new_covobs.keys())
+        else:
+            cnames_loc = cnames
+        for name in cnames_loc:
+            res[i].names.append(name)
+            res[i].shape[name] = 1
+            res[i].idl[name] = []
+        res[i]._covobs = new_covobs
+
+    if symbol:
+        for i in range(len(res)):
+            res[i].tag = symbol[i]
+            if res[i].tag == 'None':
+                res[i].tag = None
+    if not noempty:
+        _check(len(res[0].covobs.keys()) == nc)
+    if full_output:
+        retd = {}
+        tool = file_origin.get('tool', None)
+        if tool:
+            program = tool['name'] + ' ' + tool['version']
+        else:
+            program = ''
+        retd['program'] = program
+        retd['version'] = version
+        retd['who'] = file_origin['who']
+        retd['date'] = file_origin['date']
+        retd['host'] = file_origin['host']
+        retd['description'] = descriptiond
+        retd['enstags'] = enstags
+        retd['obsdata'] = res
+        return retd
+    else:
+        return res
+
+
+def read_dobs(fname, noempty=False, full_output=False, gz=True, separator_insertion=True):
+    """Import a list of Obs from an xml.gz file in the Zeuthen dobs format.
+
+    Tags are not written or recovered automatically.
+
+    Parameters
+    ----------
+    fname : str
+        Filename of the input file.
+    noemtpy : bool
+        If True, ensembles with no contribution to the Obs are not included.
+        If False, ensembles are included as written in the file.
+    full_output : bool
+        If True, a dict containing auxiliary information and the data is returned.
+        If False, only the data is returned as list.
+    gz : bool
+        If True, assumes that data is gzipped. If False, assumes XML file.
+    separatior_insertion: str, int or bool
+        str: replace all occurences of "separator_insertion" within the replica names
+        by "|%s" % (separator_insertion) when constructing the names of the replica.
+        int: Insert the separator "|" at the position given by separator_insertion.
+        True (default): separator "|" is inserted after len(ensname), assuming that the
+        ensemble name is a prefix to the replica name.
+        None or False: No separator is inserted.
+    """
+
+    if not fname.endswith('.xml') and not fname.endswith('.gz'):
+        fname += '.xml'
+    if gz:
+        if not fname.endswith('.gz'):
+            fname += '.gz'
+        with gzip.open(fname, 'r') as fin:
+            content = fin.read().decode('utf-8')
+    else:
+        if fname.endswith('.gz'):
+            warnings.warn("Trying to read from %s without unzipping!" % fname, UserWarning)
+        with open(fname, 'r', encoding='utf-8') as fin:
+            content = fin.read()
+
+    # open and read gzipped xml file
+    infile = gzip.open(fname)
+    content = infile.read()
+
+    return import_dobs_string(content, noempty, full_output, separator_insertion=separator_insertion)
+
+
+def _dobsdict_to_xmlstring(d):
+    if isinstance(d, dict):
+        iters = ''
+        for k in d:
+            if k.startswith('#value'):
+                for li in d[k]:
+                    iters += li
+                return iters + '\n'
+            elif k.startswith('#'):
+                for li in d[k]:
+                    iters += li
+                iters = '<array>\n' + iters + '<%sarray>\n' % ('/')
+                return iters
+            if isinstance(d[k], dict):
+                iters += '<%s>\n' % (k) + _dobsdict_to_xmlstring(d[k]) + '<%s%s>\n' % ('/', k)
+            elif isinstance(d[k], str):
+                if len(d[k]) > 100:
+                    iters += '<%s>\n ' % (k) + d[k] + ' \n<%s%s>\n' % ('/', k)
+                else:
+                    iters += '<%s> ' % (k) + d[k] + ' <%s%s>\n' % ('/', k)
+            elif isinstance(d[k], list):
+                tmps = ''
+                if k in ['edata', 'cdata']:
+                    for i in range(len(d[k])):
+                        tmps += '<%s>\n' % (k) + _dobsdict_to_xmlstring(d[k][i]) + '</%s>\n' % (k)
+                else:
+                    for i in range(len(d[k])):
+                        tmps += _dobsdict_to_xmlstring(d[k][i])
+                iters += tmps
+            elif isinstance(d[k], (int, float)):
+                iters += '<%s> ' % (k) + str(d[k]) + ' <%s%s>\n' % ('/', k)
+            elif not d[k]:
+                return '\n'
+            else:
+                raise Exception('Type', type(d[k]), 'not supported in export!')
+    else:
+        raise Exception('Type', type(d), 'not supported in export!')
+    return iters
+
+
+def _dobsdict_to_xmlstring_spaces(d, space='  '):
+    s = _dobsdict_to_xmlstring(d)
+    o = ''
+    c = 0
+    cm = False
+    for li in s.split('\n'):
+        if li.startswith('<%s' % ('/')):
+            c -= 1
+            cm = True
+        for i in range(c):
+            o += space
+        o += li + '\n'
+        if li.startswith('<') and not cm:
+            if not '<%s' % ('/') in li:
+                c += 1
+        cm = False
+    return o
+
+
+def create_dobs_string(obsl, name, spec='dobs v1.0', origin='', symbol=[], who=None, enstags={}):
+    """Generate the string for the export of a list of Obs or structures containing Obs
+    to a .xml.gz file according to the Zeuthen dobs format.
+
+    Tags are not written or recovered automatically. The separator |is removed from the replica names.
+
+    Parameters
+    ----------
+    obsl : list
+        List of Obs that will be exported.
+        The Obs inside a structure do not have to be defined on the same set of configurations,
+        but the storage requirement is increased, if this is not the case.
+    name : str
+        The name of the observable.
+    spec : str
+        Optional string that describes the contents of the file.
+    origin : str
+        Specify where the data has its origin.
+    symbol : list
+        A list of symbols that describe the observables to be written. May be empty.
+    who : str
+        Provide the name of the person that exports the data.
+    enstags : dict
+        Provide alternative enstag for ensembles in the form enstags = {ename: enstag}
+        Otherwise, the ensemble name is used.
+    """
+    od = {}
+    r_names = []
+    for o in obsl:
+        r_names += [name for name in o.names if name.split('|')[0] in o.mc_names]
+    r_names = sorted(set(r_names))
+    mc_names = sorted(set([n.split('|')[0] for n in r_names]))
+    for tmpname in mc_names:
+        if tmpname not in enstags:
+            enstags[tmpname] = tmpname
+    ne = len(set(mc_names))
+    cov_names = []
+    for o in obsl:
+        cov_names += list(o.cov_names)
+    cov_names = sorted(set(cov_names))
+    nc = len(set(cov_names))
+    od['OBSERVABLES'] = {}
+    od['OBSERVABLES']['SCHEMA'] = {'NAME': 'lattobs', 'VERSION': '1.0'}
+    if who is None:
+        who = getpass.getuser()
+    od['OBSERVABLES']['origin'] = {
+        'who': who,
+        'date': str(datetime.datetime.now())[:-7],
+        'host': socket.gethostname(),
+        'tool': {'name': 'pyerrors', 'version': pyerrorsversion.__version__}}
+    od['OBSERVABLES']['dobs'] = {}
+    pd = od['OBSERVABLES']['dobs']
+    pd['spec'] = spec
+    pd['origin'] = origin
+    pd['name'] = name
+    pd['array'] = {}
+    pd['array']['id'] = 'val'
+    pd['array']['layout'] = '1 f%d' % (len(obsl))
+    osymbol = ''
+    if symbol:
+        if not isinstance(symbol, list):
+            raise Exception('Symbol has to be a list!')
+        if not (len(symbol) == 0 or len(symbol) == len(obsl)):
+            raise Exception('Symbol has to be a list of lenght 0 or %d!' % (len(obsl)))
+        osymbol = symbol[0]
+        for s in symbol[1:]:
+            osymbol += ' %s' % s
+        pd['array']['symbol'] = osymbol
+
+    pd['array']['#values'] = ['  '.join(['%1.16e' % o.value for o in obsl])]
+    pd['ne'] = '%d' % (ne)
+    pd['nc'] = '%d' % (nc)
+    pd['edata'] = []
+    for name in mc_names:
+        ed = {}
+        ed['enstag'] = enstags[name]
+        onames = sorted([n for n in r_names if (n.startswith(name + '|') or n == name)])
+        nr = len(onames)
+        ed['nr'] = nr
+        ed[''] = []
+
+        for r in range(nr):
+            ad = {}
+            repname = onames[r]
+            ad['id'] = repname.replace('|', '')
+            idx = _merge_idx([o.idl.get(repname, []) for o in obsl])
+            Nconf = len(idx)
+            layout = '%d i f%d' % (Nconf, len(obsl))
+            ad['layout'] = layout
+            data = ''
+            counters = [0 for o in obsl]
+            for ci in idx:
+                data += '%d ' % ci
+                for oi in range(len(obsl)):
+                    o = obsl[oi]
+                    if repname in o.idl:
+                        if counters[oi] < 0:
+                            data += '0 '
+                            continue
+                        if o.idl[repname][counters[oi]] == ci:
+                            num = o.deltas[repname][counters[oi]]
+                            if num == 0:
+                                data += '0 '
+                            else:
+                                data += '%1.16e ' % (num)
+                            counters[oi] += 1
+                            if counters[oi] >= len(o.idl[repname]):
+                                counters[oi] = -1
+                        else:
+                            data += '0 '
+                    else:
+                        data += '0 '
+                data += '\n'
+            ad['#data'] = data
+            ed[''].append(ad)
+        pd['edata'].append(ed)
+
+        allcov = {}
+        for o in obsl:
+            for name in o.cov_names:
+                if name in allcov:
+                    if not np.array_equal(allcov[name], o.covobs[name].cov):
+                        raise Exception('Inconsistent covariance matrices for %s!' % (name))
+                else:
+                    allcov[name] = o.covobs[name].cov
+        pd['cdata'] = []
+        for name in cov_names:
+            cd = {}
+            cd['id'] = name
+
+            covd = {'id': 'cov'}
+            if allcov[name].shape == ():
+                ncov = 1
+                covd['layout'] = '1 1 f'
+                covd['#data'] = '%1.14e' % (allcov[name])
+            else:
+                shape = allcov[name].shape
+                assert (shape[0] == shape[1])
+                ncov = shape[0]
+                covd['layout'] = '%d %d f' % (ncov, ncov)
+                ds = ''
+                for i in range(ncov):
+                    for j in range(ncov):
+                        val = allcov[name][i][j]
+                        if val == 0:
+                            ds += '0 '
+                        else:
+                            ds += '%1.14e ' % (val)
+                    ds += '\n'
+                covd['#data'] = ds
+
+            gradd = {'id': 'grad'}
+            gradd['layout'] = '%d f%d' % (ncov, len(obsl))
+            ds = ''
+            for i in range(ncov):
+                for o in obsl:
+                    if name in o.covobs:
+                        val = o.covobs[name].grad[i]
+                        if val != 0:
+                            ds += '%1.14e ' % (val)
+                        else:
+                            ds += '0 '
+                    else:
+                        ds += '0 '
+            gradd['#data'] = ds
+            cd['array'] = [covd, gradd]
+            pd['cdata'].append(cd)
+
+    rs = '<?xml version="1.0" encoding="utf-8"?>\n' + _dobsdict_to_xmlstring_spaces(od)
+
+    return rs
+
+
+def write_dobs(obsl, fname, name, spec='dobs v1.0', origin='', symbol=[], who=None, enstags={}, gz=True):
+    """Export a list of Obs or structures containing Obs to a .xml.gz file
+    according to the Zeuthen dobs format.
+
+    Tags are not written or recovered automatically. The separator | is removed from the replica names.
+
+    Parameters
+    ----------
+    obsl : list
+        List of Obs that will be exported.
+        The Obs inside a structure do not have to be defined on the same set of configurations,
+        but the storage requirement is increased, if this is not the case.
+    fname : str
+        Filename of the output file.
+    name : str
+        The name of the observable.
+    spec : str
+        Optional string that describes the contents of the file.
+    origin : str
+        Specify where the data has its origin.
+    symbol : list
+        A list of symbols that describe the observables to be written. May be empty.
+    who : str
+        Provide the name of the person that exports the data.
+    enstags : dict
+        Provide alternative enstag for ensembles in the form enstags = {ename: enstag}
+        Otherwise, the ensemble name is used.
+    gz : bool
+        If True, the output is a gzipped XML. If False, the output is a XML file.
+    """
+
+    dobsstring = create_dobs_string(obsl, name, spec, origin, symbol, who, enstags=enstags)
+
+    if not fname.endswith('.xml') and not fname.endswith('.gz'):
+        fname += '.xml'
+
+    if gz:
+        if not fname.endswith('.gz'):
+            fname += '.gz'
+
+        fp = gzip.open(fname, 'wb')
+        fp.write(dobsstring.encode('utf-8'))
+    else:
+        fp = open(fname, 'w', encoding='utf-8')
+        fp.write(dobsstring)
+    fp.close()
diff --git a/setup.py b/setup.py
index b578a727..ceeb7e92 100644
--- a/setup.py
+++ b/setup.py
@@ -9,5 +9,5 @@ setup(name='pyerrors',
       author_email='fabian.joswig@ed.ac.uk',
       packages=find_packages(),
       python_requires='>=3.6.0',
-      install_requires=['numpy>=1.16', 'autograd @ git+https://github.com/HIPS/autograd.git', 'numdifftools', 'matplotlib>=3.3', 'scipy', 'iminuit>=2', 'h5py']
+      install_requires=['numpy>=1.16', 'autograd @ git+https://github.com/HIPS/autograd.git', 'numdifftools', 'matplotlib>=3.3', 'scipy', 'iminuit>=2', 'h5py', 'lxml']
      )
diff --git a/tests/json_io_test.py b/tests/json_io_test.py
index 04ccfd9f..3b312ec0 100644
--- a/tests/json_io_test.py
+++ b/tests/json_io_test.py
@@ -3,6 +3,7 @@ import gzip
 import numpy as np
 import pyerrors as pe
 import pyerrors.input.json as jsonio
+import pyerrors.input.dobs as dobsio
 import pytest
 
 
@@ -251,3 +252,98 @@ def test_renorm_deriv_of_corr(tmp_path):
     pe.input.json.dump_to_json(c, (tmp_path / 'test').as_posix())
     recover = pe.input.json.load_json((tmp_path / 'test').as_posix())
     assert np.all([o == 0 for o in (c - recover)[1:-1]])
+
+
+def test_dobsio():
+    o = pe.pseudo_Obs(1.0, .2, 'one|r1')
+    o1 = pe.pseudo_Obs(1.5, .2, 'one|r1')
+
+    ol = [o, o1]
+    fname = 'test_rw'
+    dobsio.write_pobs(ol, fname, 'Testobs')
+    rl = dobsio.read_pobs(fname)
+
+    for i in range(len(ol)):
+        assert (ol[i] - rl[i].is_zero())
+
+    od = {
+        'obsdata': ol,
+        'name': 'testn',
+        'spec': 'tests',
+        'origin': 'testo',
+        'symbol': ['A', 'B']
+    }
+    dobsio.write_pobs(ol, fname, od['name'], od['spec'], od['origin'], od['symbol'])
+    rd = dobsio.read_pobs(fname, full_output=True)
+
+    for i in range(len(od['obsdata'])):
+        assert (od['obsdata'][i] - rd['obsdata'][i].is_zero())
+
+    assert(od['spec'] == rd['description']['spec'])
+    assert(od['origin'] == rd['description']['origin'])
+    assert(od['name'] == rd['description']['name'])
+    assert(rd['description']['enstag'] == ol[0].e_names[0])
+
+    dobsio.write_dobs(ol, fname, 'Testobs')
+    rl = dobsio.read_dobs(fname)
+
+    for i in range(len(ol)):
+        assert (ol[i] - rl[i].is_zero())
+
+    dobsio.write_dobs(ol, fname, od['name'], od['spec'], od['origin'], od['symbol'])
+    rd = dobsio.read_dobs(fname, full_output=True)
+    os.remove(fname + '.xml.gz')
+
+    for i in range(len(od['obsdata'])):
+        assert (od['obsdata'][i] - rd['obsdata'][i].is_zero())
+
+    assert(od['spec'] == rd['description']['spec'])
+    assert(od['origin'] == rd['description']['origin'])
+    assert(od['name'] == rd['description']['name'])
+
+    o2 = pe.pseudo_Obs(0.5, .1, 'two|r1')
+    o3 = pe.pseudo_Obs(0.5, .1, 'two|r2')
+    o4 = pe.merge_obs([o2, o3])
+    otag = 'This has been merged!'
+    o4.tag = otag
+    do = o - .2 * o4
+    co1 = pe.cov_Obs(1., .123, 'cov1')
+    co3 = pe.cov_Obs(4., .1 ** 2, 'cov3')
+    do *= co1 / co3
+    do.tag = {'A': 2}
+
+    o5 = pe.pseudo_Obs(0.8, .1, 'two|r2')
+    co2 = pe.cov_Obs([1, 2], [[.12, .004], [.004, .02]], 'cov2')
+    o5 /= co2[0]
+    o5.tag = 2 * otag
+
+    tt1 = pe.Obs([np.random.rand(100)], ['t|r1'], idl=[range(2, 202, 2)])
+    tt2 = pe.Obs([np.random.rand(100)], ['t|r2'], idl=[range(2, 202, 2)])
+    tt3 = pe.Obs([np.random.rand(102)], ['qe|r1'])
+
+    tt = tt1 + tt2 + tt3
+
+    tt.tag = 'Test Obs: Ä'
+
+    ol = [o2, o3, o4, do, o5, tt]
+    print(ol)
+    fname = 'test_rw'
+
+    dobsio.write_dobs(ol, fname, 'TEST')
+
+    rl = dobsio.read_dobs(fname, noempty=True)
+    os.remove(fname + '.xml.gz')
+    [o.gamma_method() for o in rl]
+
+    for i in range(len(ol)):
+        assert (ol[i] - rl[i].is_zero())
+
+    for i in range(len(ol)):
+        if isinstance(ol[i], pe.Obs):
+            o = ol[i] - rl[i]
+            assert(o.is_zero())
+        or1 = np.ravel(ol[i])
+        or2 = np.ravel(rl[i])
+        for j in range(len(or1)):
+            o = or1[j] - or2[j]
+            assert(o.is_zero())