Habartice

Whose archive this ground was

Habartice sits in Silesia. Which empire held it decides which archive holds its records — and for some of these years that is not the archive you would guess.

Where the burials are

The other end of a parish: the register records the burial and the ground holds it, and a family that left no other trace very often left a stone. These are points on a map, not record collections — from Wikidata and OpenStreetMap, nobody here has visited one, and whether the burial register survives is a question for the parish or the municipality above. Grounds OpenStreetMap has not named are here too. They used to be left out, on the reasoning that a shape with no name is not a place anyone can ask about — which cost Senj its own graveyard, 2 km away and unnamed, while a named one 7 km off was all this page would show. Being unnamed is a fact about the map data, not about the ground. Only inside a mapped record region.

• A burial ground ↗ unnamed in OpenStreetMap 0.5 km OpenStreetMap
• A burial ground ↗ unnamed in OpenStreetMap 1.2 km OpenStreetMap
• A burial ground ↗ unnamed in OpenStreetMap 2.5 km OpenStreetMap
• A burial ground ↗ unnamed in OpenStreetMap 4.6 km OpenStreetMap
• A burial ground ↗ unnamed in OpenStreetMap 5 km OpenStreetMap

3 volumes catalogued here

The level below «a collection covers this place»: which book, for which years, and the link that opens its images. From FamilySearch's own catalogue. Nobody here has opened any of them. You can, though — mark the pages you have searched on any row below and this browser will remember where you stopped.

National sources for Poland, unverified

From Wikidata, CC0 — the national archive and the national library where the country has one. Nobody here has opened these. The link answered when it was asked and that is the whole of what this map knows about them. They are listed because a country with no named source at all is the worse answer: 199 of the 237 countries here had nothing but the worldwide services, which cover everywhere and therefore say nothing about anywhere.

• National Archives in Kraków ↗ national archive answered with an error
• National Digital Archives ↗ national archive
• Polish Central Archives of Historical Records ↗ national archive did not answer this checker
• National Library of Poland ↗ national library

The sources for Poland

National and regional sources that cover this ground whether or not anybody has walked it here — which makes them the answer, not the consolation, for a place with no volumes recorded on it.

• Szukaj w Archiwach Images, free, no account Warsaw, Poland
  The Polish state archives' scans, free and without an account. Covers ground now in Ukraine, Belarus and Lithuania as well as Poland. *** BUT IT IS BOT-GATED AND A SCRIPT CANNOT READ IT. *** On 11 October 2026 every request to szukajwarchiwach.gov.pl — the search API and the plain search page alike — returned a 212-byte Incapsula challenge page with “robots: noindex,nofollow” and nothing else. FREE TO A PERSON IN A BROWSER, CLOSED TO ANYTHING AUTOMATED, and this atlas records that rather than working around it. It matters because this is the national catalogue: it is the instrument that answers “does this fonds exist and where is it”, which the sibling portal Genealogia w Archiwach cannot do, so a question that ought to take one query becomes a letter to an archive.
• Archiwum Główne Akt Dawnych (AGAD) A catalogue entry only — the paper is not photographed Warsaw, Poland
  Warsaw's archive of the old Republic — the metrical books of the Polish partitions, including eastern Galician parishes now in Ukraine. Catalogued here; the scans are published through Szukaj w Archiwach.
• Bundesarchiv A catalogue entry only — the paper is not photographed Koblenz, Germany
  The German Federal Archives. Civil and military administration of the Prussian provinces, the wartime records, and the card indexes of the expelled — much of it for ground that is no longer German.
• Matricula Online Images, free, no account Passau, Germany
  Diocesan parish registers photographed and served free, without an account. Austria and much of Germany and Poland; the Italian entries are South Tyrol.
• Archiwum Państwowe w Bydgoszczy Nobody has walked this one yet Bydgoszcz, Poland
  The Polish state archive for the Kuyavian-Pomeranian voivodeship, holding the civil registers of the former Prussian registry offices of the region, Szubin among them. IT ANSWERS LETTERS AND WILL SHARE SCANS: three e-mails in September 2026 were each answered within about eight days, in Polish, over the Director's signature with a named case officer. It points first at genealogiawarchiwach.pl for what is already digitised, and then supplied a 103-scan volume that is NOT on that portal, shared from its own OneDrive. TWO HOLDINGS FACTS IT STATED BY LETTER THAT NO ONLINE CATALOGUE SHOWS: the metrical books of the Jewish religious community of Szubin survive ONLY for 1820-1847; and the alphabetical death indexes (skorowidze zgonów) of USC Szubin Miasto survive ONLY for 1874-1878 and 1894-1924, with nothing between. An archivist there will also read a difficult line from a register on request and say plainly where the reading is uncertain.
• GenTeam.at Images, free, account needed Vienna, Austria
  Volunteer indexes to the Habsburg lands — Vienna's parish registers, Jewish records, school and guild lists. Free with an account, and it reaches ground the church portals do not.
• Geneteka An index only — no images Warsaw, Poland
  A volunteer index of Polish parish registers, by surname and parish. An index only — it points you at the scan rather than showing it. IT IS UP AND WORKING AS OF 11 OCTOBER 2026 (it was answering HTTP 500 on 16 September). HOW TO QUERY IT. The HTML page at index.php returns the form and an EMPTY TABLE; the rows come from api/getAct.php as JSON, taking the same query string plus DataTables' draw/start/length, and LENGTH IS CAPPED AT FIFTY however much is asked for. api/getParish.php?search=<place> answers whether a place is indexed at all, and that is the first call worth making — it is cheaper than any search and it is usually the real answer. THE PROVINCE CODES ARE POSITIONAL (01ds, 02kp, 03lb, 04ls, 05ld, 06mp, 07mz, 08op, 09pk, 10pl, 11pm, 12sl, 13sk, 14wm, 15wp, 16zp) AND AN INVALID CODE FALLS BACK SILENTLY TO 01ds — so every query runs against dolnośląskie and returns nothing, INCLUDING THE CONTROL. Always run a common-surname control; a broken query and an empty province look identical. TWO COVERAGE FACTS MEASURED IN USE, BOTH OF WHICH ARE REASONS A ZERO HERE MAY MEAN NOTHING. (1) THE INDEX IS OF CATHOLIC AND PROTESTANT PARISH BOOKS AND HOLDS ALMOST NO JEWISH RECORDS. In pomorskie, births, exact match: SCHMIDT 1,134, KOWALSKI 1,023, MÜLLER 767 — against COHN 1, LEVY 0, LEVIN 0, ISRAEL 1, JACOBSOHN 2, ROSENTHAL 3. Four figures against single digits. A Jewish surname's absence is the index speaking, not the register, and should never be recorded as a negative without that control. (2) COVERAGE IS BY PARISH AND IS WILDLY UNEVEN, SO ASK THE PLACE QUESTION FIRST. In the former German districts of pomorskie, nine of eleven towns tested were not indexed at all, Człuchów (Schlochau) and Czarne (Hammerstein) among them — and a place search on a German town's modern Polish name can return a DIFFERENT town of that name in another province, which is a trap. Of the two that were indexed, one stops before 1800 and one before 1840, while the same province indexes 1874-1925 well elsewhere. In warmińsko-mazurskie the same shape appeared: 278 parishes including Olsztyn and Reszeł and their civil registries, and not Bössau, Biesowo, Jeziorany or Zerbuń.
• Genealogia w Archiwach Images, free, no account Poznań, Poland
  A Polish state-archives portal of scanned metrical books — civil registration from the Prussian partition (1874 on), duplicates of Roman Catholic parish registers from before that, and population and registration books from 1837. Separate from Szukaj w Archiwach and run by its own archives. Secondary sources say the scans are free and need no account; NOBODY HERE HAS CONFIRMED THAT, because the site is a Vaadin application that renders nothing without JavaScript, so it is marked unsurveyed rather than taking a blog's word for what it costs. SCANS ARE SERVED BY A LORIS IIIF IMAGE SERVER and can be read in place without downloading: /loris/<id>/<x,y,w,h>/<width>,/0/default.jpg returns an arbitrary region at up to TWICE native resolution, which is enough to settle a difficult hand. info.json returns 403 while image requests succeed, and anything above 2x returns a bare 404. The viewer's own canvas is tainted, so the tile server is the route and screenshots are not. THE PORTAL WILL NOT OPEN A UNIT BY ITS SIGNATURE: a scan's modal id resolves only inside the current search result set, so a neighbouring scan cannot be reached by guessing an id. The route that works is to open any scan's modal, click 'View the document' to expose the unit's own scan list, click Show on the scan wanted, and take the new Loris id from the network log. *** SURVEYED 11 OCTOBER 2026, AND IT IS FREE AND NEEDS NO ACCOUNT. *** Two sessions of this archive have read register images off it without signing in or paying anything, so the secondary sources are right. HOW IT ACTUALLY WORKS. The scans are served by a LORIS IIIF IMAGE SERVER and arbitrary regions can be pulled by pixel coordinates at up to TWICE native resolution (/loris/<id>/<x,y,w,h>/<width>,/0/default.jpg) — the viewer's own canvas is tainted and cannot be copied, the tile server can, and screenshotting the zoomed viewer is the slow wrong road. THREE TRAPS. (1) THE DEFAULT SEARCH RUNS IN SCAN MODE over volunteer-indexed PERSON names, not over the catalogue, so it cannot answer “does fonds X exist” — a search for a town plus a record type returns zero while the town's registers sit on the portal. Use “Browse Teams (PRADZIAD)” for the fonds. (2) IT IS A VAADIN APPLICATION AND THE URL HASH DOES NOT DRIVE IT: pasting a query into the address bar re-renders the previous result, and the form has to be driven. (3) On 11 October 2026 the portal was displaying ITS OWN NOTICE of difficulties while work proceeds on reconfiguring its internet links, and the fonds browser would not open — so a zero from this portal should be checked against the notice before it is believed.

Sources that cover everywhere

Not about Poland in particular — these reach most of the world, which includes here. They are listed after the local sources because a body that holds this ground's own paper is a better first stop than one that holds a little of everywhere's. Worth knowing they exist all the same.

• OldNews.com Nobody has walked this one yet
  Digitised historical newspapers, indexed by town. The CATALOGUE is harvested and published here — 38,151 papers in 6,450 towns across 75 countries, of which 5,968 towns join to a point on this map — because their robots.txt says Allow: /en/newspapers/* and the whole structure is in the URL. WHAT IT COSTS IS NOT KNOWN, and that is why this says unsurveyed rather than free or paid: the same robots.txt disallows /*/record and /*/search, so the pages that would answer it are the pages not to read. There is a log-in; whether an image is free behind it has not been established, and a colour on the map is a promise this cannot keep yet.
• Internet Archive Images, free, no account
  Full-text search across millions of digitised books, including parish histories, nobiliaries and local monographs that name ordinary people. THREE THINGS VERIFIED 2026-10-11 WHILE REBUILDING A FAMILY'S 18th-CENTURY LINE FROM IT. (1) SEARCH THE RAREST TOKEN, NOT THE SURNAME. A search on one unusual middle name returned 329 hits and pulled out a 1972 family memoir, a reference work of colonial governors, a Falklands bibliography, an 1866 genealogical query and two monthly Army Lists — none of which a surname search had ever surfaced, because the book was CATALOGUED UNDER A MANGLED FORM OF THE SURNAME. (2) THE FULL-TEXT SEARCH HAS A PLAIN JSON API, which the site's own search page calls and which returns highlighted snippets: /services/search/beta/page_production/?service_backend=fts&user_query=%22PHRASE%22&hits_per_page=100&page=1&aggregations=false. It needs an ordinary browser User-Agent AND an Accept header; without the Accept header the site answers some requests with a flat 404. (3) FOR ANY OUT-OF-COPYRIGHT ITEM THE WHOLE TEXT IS ONE FILE — archive.org/download/<identifier>/<identifier>_djvu.txt — so a whole volume can be searched locally rather than page by page. Four Army Lists of 1775-1783, about 400KB of text each, were downloaded and searched entire in under a minute; that is how an absence can be MEASURED, by counting what the volume does contain. IN-COPYRIGHT ITEMS ARE LENDING-RESTRICTED: the item page still shows the metadata and the LIST OF ILLUSTRATIONS without borrowing, which was enough to recover a woman's missing middle name, but search-inside and the text need an account.
• Arolsen Archives Images, free, no account
  The former International Tracing Service. Concentration camp, forced labour and displaced-persons records for some seventeen and a half million people, free and fully indexed.
• Find a Grave Images, free, no account
  Volunteer-photographed headstones and cemetery indexes, worldwide, with very uneven coverage. Owned by Ancestry but free to read.
• WikiTree Images, free, no account
  A free, single shared tree with an unusually strong sourcing culture. Not a record source.
• Yad Vashem — Central Database of Shoah Victims' Names Images, free, no account
  Names and testimony pages for victims of the Shoah, free, no account. THE SEARCH IS FULLY URL-DRIVEN even though the site is an Angular app: /en/names/search-results-names?page=N&gn_last_name=X&gnt_last_name=yvSynonym&gn_first_name=Y&gnt_first_name=yvSynonym, with gnt_ taking yvSynonym, literal or soundex. THE RESULTS COME IN THREE TABS AND THE SECOND IS THE ONE ALMOST NOBODY USES: “Results Regarding Victims”, “RESULTS REGARDING DATA SUBMITTERS” (append #submitter to the URL) and “Additional Results”. The submitter tab lists every record whose SUBMITTER matched the name searched — so SEARCHING A SURVIVOR'S NAME THERE RETURNS THE DEAD HE WROTE DOWN, and one Page of Testimony can be turned into a whole family in a single query. There are also Submitter Last Name and Submitter First Name fields in the advanced form for the same purpose. TWO THINGS MEASURED IN USE. (1) THE SYNONYM EXPANSION IS LOOSE and will fold in a different submitter of a similar name from another country — in one test, 24 of 29 results belonged to somebody else and had to be separated by place — so a submitter set must be READ and not counted. (2) A PAGE OF TESTIMONY IS A RELATIONSHIP DOCUMENT, NOT A DATE DOCUMENT. Written from memory, often decades later and often in Hebrew and auto-translated back, it is the best evidence there is for who was whose son and the worst for when: in one family read here a Page disagreed with the German federal memorial list on a birth day AND a death year (the lists were right both times), two Pages by the same man for the same victim gave two different places of death, and the same parents appeared under German forenames on one form and Hebrew ones on the other — which is itself a gift, because it hands you the Hebrew names a community register would use. What a Page gives that no list gives is the single word in the “relationship to victim” field: son, nephew, grandson.
• FamilySearch Images, free, account needed
  The largest free collection of record images and indexes in the world. A free account is required, and a minority of images are restricted to Family History Centres and affiliate libraries. TWO TRAPS IN THE CROATIAN HOLDINGS, both verified by use on 2026-10-09 and both capable of producing a confident false nil. (1) THE SAME PARISH SITS IN MORE THAN ONE COLLECTION. Bakar (Buccari) records are indexed under BOTH 'Croatia, Church Books, 1516-1994' (cc=2040054) AND 'Croatia, Delnice Deanery Catholic Church Books, 1571-1926'. A search restricted to 2040054 returns ZERO for records that demonstrably exist in the Delnice Deanery set - a marriage of 1783 at Bakar was found only once the collection filter was dropped. So a nil reported against one Croatian collection is a statement about that collection, never about the parish; and the catalogue's film list for a place may be split across collections that do not cross-reference each other. (2) SEARCH URL PARAMETERS SILENTLY FAIL. Hand-built query strings of the form /search/record/results?q.surname=X&q.anyPlace=Y returned 'Historical Record Search Results (0)' for a record whose own detail page gives exactly that surname and exactly that place string. The same query typed into the search form returns it. Place and collection parameters are the ones observed to fail; a bare q.surname + q.givenName pair did work. THE PRACTICAL RULE: never report a FamilySearch nil from a URL you assembled. Drive the form, and run a known-positive control - an ark id you already hold - through the same path first. An index entry is also a pointer and not a reading: the 1783 marriage above is indexed as two names and a date, and the words 'ex Gollogorizza' in the entry itself - the words that place the groom 130 km away in another crown land and make the record foreign to the Bakar set it sits in - appear nowhere in the index. Open the image before building or dismissing a household on an indexed record. Film and image-group identifiers DO work as stable handles where this collection's viewer carries no wc= waypoint: the Bakar marriage volume is film 005494714, 'Bakar. Religious Marriage Registers 1739-1808', item 1, 226 images, image group M99F-VRM. FULL-RESOLUTION IMAGES CAN BE DOWNLOADED, and it is worth knowing because the viewer's own zoom is a poor substitute for cropping locally. The download control in the image viewer opens a dialog offering 'PDF Record Image with Highlights', 'PDF with No Highlights', 'JPG Only' and 'No Record Image'. Choose JPG Only: it delivers the plate at native scan resolution - a Croatian parish register leaf came down at 5975x4206 - named <filmnumber>_<imagenumber>.jpeg, e.g. 005494714_00172.jpeg, which is a self-describing filename worth keeping. Note that the DeepZoom tiles the viewer itself loads (service/records/storage/deepzoomcloud/dz/v1/apid:<APID>/image_files/<level>/<col>_<row>.jpg) return 401 to an unauthenticated fetch and are blocked by CORS from the www origin, so scraping them is neither possible nor necessary; the download button is the supported route and it works. Resolution matters for more than comfort: a contested initial letter in a surname was settled only by cropping two words at identical magnification from the downloaded file, and the same comparison attempted through the viewer's zoom had been inconclusive.
• Geni Images, free, account needed
  One shared world tree. Useful for finding other researchers on the same line; not a record source, and its assertions are only as good as their citations.
• JewishGen Images, free, account needed
  The central Jewish genealogy resource — town-level KehilaLinks, burial and Holocaust databases, and revision lists for eastern Europe.
• BillionGraves Partly free, partly paid
  GPS-tagged headstone photographs. Smaller than Find a Grave but its coordinates are per-stone, which Find a Grave's often are not. FREE AND PAID ARE SPLIT DOWN THE MIDDLE, surveyed 7 October 2026 at a South African cemetery: the search, the transcription — name with full birth and death dates — and the per-stone GPS coordinates are free and need no account, and so is a 180x135 thumbnail of the stone. THE FULL-RESOLUTION PHOTOGRAPH IS NOT: it requires an account and a subscription. The thumbnail is far too small to read an inscription, so a transcription taken from here CANNOT BE CHECKED AGAINST THE STONE without paying — which matters, because a volunteer's reading of a weathered date is exactly the thing an archive will want to verify. AND ITS SEARCH URL SILENTLY IGNORES WHAT YOU ASK IT, which makes it USABLE FOR A POSITIVE AND USELESS FOR A NEGATIVE: a hit here is a hit, and no absence may be quoted from it. Verified 9 October: `cemetery_country=United+Kingdom` is accepted and disregarded — the results still carry United States and Australian stones — and `page` does nothing at all, `page=7` returning the identical six records as `page=1` under a footer reading «Page 1 of 178». The footer asserts a depth the URL cannot reach, so the first page is not the result set and nothing that looks like «no match in that country» is one.
• Geneanet Partly free, partly paid
  Member-contributed trees and transcriptions, strongest in France. Much is free; a Premium tier gates some search and images.
• Ancestry Images, behind a paywall
  The largest commercial collection. Strong on the English-speaking world, censuses and immigration; weak on continental parish registers.
• MyHeritage Images, behind a paywall
  Commercial, with a large tree network and good European newspaper and directory coverage. Search is free; opening a record is not. A LINK CHECKER CANNOT TELL YOU WHETHER A MYHERITAGE CITATION IS ALIVE, measured 2026-10-09 against 428 of them carried into a family archive from a GEDCOM export. Record pages answer an unauthenticated request in one of two ways, and NEITHER IS AN HONEST 200: either a 302 to a door, or HTTP 200 WITH AN 843-BYTE BODY READING «Request unsuccessful. Incapsula incident ID: …». A checker that reads the status code alone therefore reports every one of them as RESOLVING. Any archive auditing its outward citations has to read the body, and should grade these neither alive nor dead.

FamilySearch collections that reach here

A collection is a claim over an area, not a book standing in this town — so these are ranked by how specifically they reach Habartice. A free account is needed to open the images, and some are restricted to a Family History Centre.