Terms and privacy

This is the whole of it. There is no second, longer agreement behind this page and no clause you have not read. One person runs this archive, it is not a business, and the useful thing is to write down what actually happens rather than what a form would say happens.

Who is responsible

I am Michael Bommarito. I built this archive and I run it, alone, at michael.bommarito@gmail.com. More about me is on my own site, and more about the archive's debts is on the credits page.

I make no money from this. Nothing here is for sale and nothing is advertised. That matters more than it sounds: it is the condition on which most of the material below is published at all, and it is written into the archive's own rights files in those words.

Using what is here

What you may republish depends on the item, and the answer is on the rights and reuse page, which is the authority. In short: rights are recorded per file in the holder's own words, publication is denied by default, federal public-domain products are free to reproduce, and material drawn from OpenStreetMap carries an attribution condition. Reading it, quoting it and linking to it need no permission from me, and never will.

Every page here has a stable address and is meant to keep it. Link to anything.

What is not checked

6,186 of the 6,373 media works here carry a description and a set of tags that a machine wrote. A model was shown the image and the catalogue record and asked what it saw. Nobody checked the answer. Every one of them is labelled on the page as machine-written and unchecked, and they are wrong often enough to matter — about who is pictured, where, when, and what is happening. Treat them as a way of finding an image, not as a statement about it.

The same caution applies more widely. The full text of the newspapers, books and papers is uncorrected OCR, and it misreads names constantly. The encyclopedia articles cite a source for every claim, which is a different and lesser thing than being verified: a citation tells you where to go and check, and going to check is the point. Where an item's dating or identification is uncertain, the page says so.

So: I do not promise that anything here is accurate, complete or current. If a decision of yours turns on something you read here — a boundary, a date, a name, a deed, a depth on a chart — go to the source the page cites and read it there. Every page names its source and links to it, and that is what the link is for.

Corrections are wanted, and the machine-written descriptions are the part where somebody who recognises a face or a shoreline knows more than the model did. The corrections page says how to send one.

If you are in a photograph

2,859 of the photographs here came from public Facebook groups and pages run by island organisations and residents. Their rights are recorded, verbatim, as community source with no stated terms, published non-commercially with attribution to the poster and a link back to the original post. That is the basis for publishing them. It is not a claim that anyone signed anything over, and it is not an argument I intend to win against somebody who would rather not be there.

If you are in a photograph here, or you posted one, you can have it removed or altered. You do not need a reason and you will not be asked for one. Write to michael.bommarito@gmail.com; the corrections page says what to include and what happens next. The image stops being published, usually the same day, and everything written from looking at it goes too: the description, the tags, the transcription. The record keeps its address and says plainly that the image was withdrawn, so that links to it do not break and the archive can account for its own gaps.

What I cannot honestly promise is that it is as if the image had never been here. The withdrawal is written into a register file that I keep under version control, so the fact of it — the work's id, the date, and one public sentence of reason — stays in that file's history where I can see it. That is deliberate: an archive that silently shrinks cannot be audited, including by me a year later wondering about a gap. Who asked, and how, goes in a private note in the same file that is never published. Neither repository is public today. Beyond that, I have no reach at all: if someone else already copied the image, or a search engine or the Internet Archive cached the page, taking it down here does not take it down there, and I will not pretend otherwise.

What this site learns about you

Reading here requires no account and no consent dialogue. Every page was built before you asked for it and is served as a file. Below is everything, and it was checked against the code rather than remembered.

The host. Pages are served by AWS Amplify. Like every web server on earth it receives the request itself — your IP address, the page you asked for, your browser's user-agent string, and the referring page if there was one. That happens at the host, in nothing I wrote, and what Amazon retains is governed by Amazon's terms rather than mine. I run no logging of my own on top of it.

Analytics, which this site does use. One third-party script loads on every page: Plausible. Your browser fetches it from plausible.io, a server I do not run, and that request carries your IP address, the page URL, the referrer and your user agent. What Plausible does with that is their statement to make and not mine, so read it: their data policy says it sets no cookies, collects no personal data and creates no persistent or cross-site identifier. What comes back to me is counts — how many people read a page, which pages, which country, what linked here. Nothing in it identifies a reader, and nothing in it would let me learn that a particular person read a particular record. If you would rather it did not load, block it; the site is built entirely without JavaScript and loses nothing at all.

Searching, which works two different ways. The encyclopedia search is Pagefind. Its index is built when the site is built and served from this domain as static files, and the searching runs in your browser: those queries never leave your device, and I could not see them if I wanted to. The archive search — the full text of the newspapers, books and papers — cannot work that way, because the index is far too large to send to a browser. That tab posts your query to search.beaverislandarchive.org, which is my own server and not a third party's, and those queries do leave your device. I have built no query logging into the site, but there is a web server in front of the index and I am not going to promise you it keeps no record of a request. The two tabs are labelled on the search page; the distinction is real and this is the honest version of it.

Storage in your browser. This site sets no cookies. It stores exactly one thing, under the key galview, holding the word sheet or rows: whether you last looked at the media gallery as a grid or as rows. It never leaves your browser and identifies nothing. Clearing your site data deletes it. There is no other storage and no fingerprinting — no record of what you have already looked at.

Two places where something loads from elsewhere. On the article pages for the 1938 Lomax recordings, the illustration is loaded directly from the Library of Congress's own image server rather than copied here, so those pages make a request to loc.gov. One article embeds a video its publisher hosts on YouTube, loaded from youtube-nocookie.com. Everywhere else — the fonts, the images, the search index, the stylesheets — is served from this domain.

If you write to me. Then I have your email address and whatever you put in the message, for as long as the message sits in my mail. I do not have a mailing list to add you to.

What I do not promise

The site may change, and pages will be corrected, rewritten and occasionally withdrawn; that is what an archive that admits its errors looks like. I promise no uptime and owe no support. A service this depends on may stop working, and then it will not work. Nothing here is legal, surveying or navigational advice. The historical charts in particular are historical: do not navigate by them.

What follows is in the standard form, in capitals, because that form is what makes it work. A disclaimer of the implied warranties has to name them and has to be conspicuous, or a court will read it as decoration and ignore it. I would rather write it plainly than have it fail quietly, so it is set apart and it is loud.

THE ARCHIVE AND EVERYTHING IN IT ARE PROVIDED "AS IS" AND "AS AVAILABLE", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED. I SPECIFICALLY DISCLAIM THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE, AND NON-INFRINGEMENT, AND ANY WARRANTY THAT THE MATERIAL IS ACCURATE, COMPLETE, CURRENT, OR UNINTERRUPTED. TO THE FULLEST EXTENT PERMITTED BY LAW, I AM NOT LIABLE FOR ANY DAMAGES OF ANY KIND ARISING FROM YOUR USE OF THIS SITE OR YOUR RELIANCE ON ANYTHING IN IT, INCLUDING DIRECT, INDIRECT, INCIDENTAL, CONSEQUENTIAL, SPECIAL, OR EXEMPLARY DAMAGES, EVEN IF I HAVE BEEN TOLD SUCH DAMAGES ARE POSSIBLE. YOUR ONLY REMEDY IS TO STOP USING THE SITE.

Some places do not allow a disclaimer that broad, and where that is so it applies as far as the law there allows and no further.

I am not going to name a court and a body of law, because I do not know what would apply to you and would rather say nothing than assert something I cannot support. The rest is plain enough. This is published in good faith, for free, by one person, with its errors visible and its sources named. If it is not worth your trust, do not rely on it — and if you can show me where it is wrong, I would rather know.

If any of this is wrong

This page describes how the site behaves, and a site changes. If you find something here that the site does not actually do, or something the site does that this page does not mention — a tracker, a request going somewhere it should not, anything stored that is not galview — that is a defect and I want to hear about it. Write to michael.bommarito@gmail.com and it will be fixed or written down.

The figures on this page are computed at build time from the archive's own media manifest. Everything stated about this website's behaviour was verified against its source and its built output on 5 August 2026.