How we research and price the coins on this site
Every mintage, rarity tier, and price range published on Collectors Key traces back to a named source. This page describes which sources we use, how we cross-check them, and what we deliberately do not claim. If a number on a coin page does not match what you find elsewhere, the methodology below should explain why. Most often the difference comes down to one of three things: we separated proof figures from circulation totals, we weighted recent auction records over older catalog values, or we treated population data as a better measure of scarcity than published mintage.
Mintage figures
The first source for any mintage is the United States Mint Director's Annual Report, which records year-by-year delivery counts for each branch and denomination from 1794 forward. For circulation strikes those numbers are usually authoritative, and we cite them directly for output from Philadelphia, Charlotte, Dahlonega, San Francisco, Carson City, New Orleans, Denver, and West Point as applicable to each series. The digitized Mint Reports hosted by the Newman Numismatic Portal have made several formerly-disputed figures definitive over the past decade.
Two cases push us off the Mint Report. The first is pre-1858 proof coinage. The Mint did not formally track proof deliveries until 1858, and the early proof-program years before 1860 are particularly thin in the archival record. For those entries we cite surviving-population estimates from Q. David Bowers, David W. Akers, and Walter Breen, three working specialists whose census work for early federal gold remains the standard reference. We say "estimated 10 to 15 known" rather than fabricate a Mint figure that does not exist.
The second case is the recurring conflation of proof and circulation totals on legacy catalog references. The 1863 Indian Princess gold dollar is a clean example: the Mint struck 6,200 circulation pieces and 50 proofs that year, but many published references show 6,250 as a single combined figure. We separate them. Where a live page on the site previously displayed the combined number, we have corrected it during the most recent series audit. The same conflation pattern recurs across most low-mintage Philadelphia gold from the 1860s and 1870s, and we have systematically split it on every series we have audited.
Rarity classification
We tier coins into four categories: Regular, Variety, Semi-Key, and Key. The tier is a population judgment, not a mintage cutoff. A 13,000-piece mintage with 800 surviving examples reads differently from a 13,000-piece mintage with 90 survivors, and the badge has to reflect that difference rather than the original delivery figure.
Survival data comes from two primary sources: the PCGS Population Report and the NGC Census. Both list how many graded examples each grading service has certified across all numerical grades, broken out by date, mintmark, and variety. We use them in combination because some collectors prefer one service over the other, and certain dates show meaningful skew between the two populations.
Population reports overcount somewhat. The same coin can be cracked out of one holder and resubmitted multiple times, generating duplicate population entries. We treat raw census numbers as a ceiling rather than an exact survival count, and for early federal gold and other denominations where named cabinets track named pieces, we cross-reference against Bowers' census work and major auction-archive concentrations to keep the estimate honest.
The Key tier is reserved for the lowest-mintage anchor of a series with verified low survival. The 1875 Indian Princess gold dollar (420 struck, perhaps 200 surviving across all grades and strike types) is a Key. The 1916-D Mercury dime is a Key. The 1909-S VDB Lincoln cent is a Key. We do not award Key status to coins simply because they are expensive. The badge reflects rarity, not market price.
Price ranges
Prices on this site are ranges, not single figures. A coin in About Uncirculated condition does not transact at one price; it transacts inside a band that depends on eye appeal, certification, recent demand, and whether the seller is a major auction house, an established dealer, or a private collector. Publishing a single number would imply a precision the market does not support.
Our primary anchor is realized auction prices from Heritage Auctions and Stack's Bowers Galleries, the two largest United States numismatic auction firms by volume and catalog reach. Realized prices are what coins actually sold for, which is the only objective measure of market value. We pull the most recent 24 to 36 months of sales for each grade tier and weight recent comparables more heavily than older ones. For thinly-traded dates we extend the window further back rather than guess.
Two secondary sources fill in where auction data is sparse. The PCGS CoinFacts price guide gives us a published asking-price reference, useful for comparison and for dates that rarely appear at auction. Established dealer asking prices from firms with public inventories provide a third reference point. We treat asking prices as ceiling indicators rather than transaction values, and we say so when an entry leans on them rather than on realized sales.
We list condition tiers separately because price scales nonlinearly with grade. The 1882 Indian Princess gold dollar in Very Fine might trade in the high three figures while an MS-65 example crosses five figures, and the gap between an MS-63 and an MS-65 of the same date can exceed the entire VF-to-AU range. The tiered breakdown is the only way to give a useful answer to "what is this coin worth."
For varieties (date logotype variants, doubled dies, repunched mintmarks, overdates), we publish premiums where the variety draws a documented market premium against the base date. Where the variety is collected but does not consistently command a premium, we say so rather than invent one.
Authentication diagnostics
When a coin entry includes specific authentication content, the diagnostics come from the same authentication community that grades and authenticates coins commercially. The PCGS counterfeit detection guides, the NGC Variety Plus references, the American Numismatic Association's Counterfeit Detection volumes, and Bowers' standard works on early federal gold (especially his references on the 1796 quarter eagle Stars variety, the 1822 half eagle, and the early dollar issues) underpin most of what we publish on counterfeit identification. For proof gold, John Dannreuther's reference work on United States proof coins is the standard source for delivery figures and surface diagnostics.
The diagnostics themselves are concrete: the 1.672 gram weight of a Type 3 gold dollar, the 26.73 gram weight of a Morgan dollar, the reeded edge versus the lettered edge on early federal gold, the 180-degree coin alignment versus the 0-degree medal alignment on certain reverses, the position of designer initials inside truncations and curls, and the specific weak-spot regions cast counterfeits tend to leave at each denomination. A digital caliper, a precise scale, and a 10x loupe will catch the great majority of casual counterfeits before money changes hands.
For high-rarity issues where every example carries a documented pedigree, provenance functions as part of the authentication chain. A previously-unknown 1804 dollar or an 1870-S three-dollar gold piece does not exist independently of its history, and the major grading services accept these only with cabinet history that traces back through known auction records. We name the relevant pedigree expectation in entries where this matters.
We do not tell collectors to "consult an expert" instead of giving them the test. The test, run with the right instruments, is the expert's first move; collectors deserve the same starting point.
What we do not do
We do not predict prices. Coin markets respond to bullion movements, auction calendars, dealer inventories, and collector demographics in ways no one forecasts reliably, and we do not pretend otherwise. We publish what coins recently sold for, with context for the conditions that produced those prices.
We do not certify or authenticate coins. Third-party grading services do that work, and a certified slab is the only reliable way to establish grade and authenticity for purposes of resale. We provide diagnostic information collectors can use to evaluate raw coins. We do not issue grading opinions.
We do not buy or sell coins. The site has no commercial connection to any auction house, dealer, or grading service. Our incentive is editorial accuracy because that is what brings collectors back to the site, and pushing transactional inventory is not part of the model.
We do not guarantee exact mintages where the historical record is incomplete. When we publish an estimate, we say so, and we name the working specialists whose population work supports the figure.
How we update entries
Mintage figures update when better primary sources surface. The Newman Numismatic Portal's ongoing digitization of Mint Director's reports, auction catalogs, and contemporary numismatic correspondence has clarified several formerly-disputed figures, and we adjust accordingly.
Rarity tiers shift when long-running survival data shifts. PCGS and NGC release fresh population reports continuously, and a date that has crossed the line from Semi-Key into Regular over the last decade because more pieces surfaced and graded gets reclassified. The reverse also happens: dates once treated as routine occasionally tighten into Semi-Key territory as graded populations stagnate against rising collector demand.
Price ranges refresh against new auction data as it accumulates. Heritage and Stack's Bowers run hundreds of sessions per year between them, and a single major auction can move the market reference for a date by a meaningful margin. We rebuild ranges as fresh comparables warrant, with attention to whether a single anomalous result actually reflects a market shift or just a one-off bidding war.
Each coin page carries a "Last edited" date in the About This Coin section, and that stamp is the canonical signal of when our position on a given mintage, classification, or price range last moved. We do not retain a public log of prior figures, so the date stamp is the way to tell whether the entry you are reading reflects the most recent audit pass on that series or a figure that has been revised since.