There was no single bitcoin price for an early exchange to copy. Each venue matched the orders in front of it, in a particular currency, through payment routes that could take longer than the price difference lasted.
No Peg and No Central Order Book
The European Central Bank described the mechanism plainly in its October 2012 report: bitcoin was not pegged to a real-world currency, and its exchange rate was determined by supply and demand. The same report listed numerous platforms operating at the time. Those facts belong together. Supply and demand did not meet in one universal queue; they met separately on Mt. Gox, TradeHill, BitMarket.eu and other venues.
BitMarket.eu divided trading again across ten currency pairs. A bid in euros did not fill an ask in dollars, and an order in rand could remain untouched while a sterling market moved. Each book generated its own best bid, best ask and last trade from the orders actually present there.
A displayed rate was therefore local evidence. It showed where one venue's participants had most recently agreed, not a binding conversion rate for the network.
Why the Differences Persisted
A price gap invites a simple trade: buy on the cheaper exchange, move the bitcoin, and sell on the more expensive one. The sequence is simple only if money and bitcoin are already positioned on both venues. Otherwise, a trader must deposit fiat, wait for bank settlement, buy, wait for blockchain confirmation, transfer to the second venue and sell before the gap closes.
Fees, confirmation requirements, withdrawal limits and banking hours all reduce the apparent spread. Counterparty settlement added another delay on venues where buyers and sellers transferred fiat directly. A quoted premium could be compensation for a slow or difficult payment route rather than a free profit.
Thin books magnified the effect. The best displayed price might cover only a small order. Buying more consumed progressively higher asks, while selling elsewhere consumed lower bids. The executable spread was smaller than the two headline numbers suggested.
The return leg is easy to forget. A completed arbitrage leaves the trader holding fiat on the expensive venue and nothing on the cheap one, so repeating the trade means sending money back through the same banking route that made the first leg slow. Capital parked on an exchange to avoid that wait was capital exposed to the exchange. Traders who kept balances on both sides could act on a gap immediately, and they carried the failure risk of both venues for the whole time they were waiting for one.
Users Compared Venues in Public
The BitMarket.eu forum thread preserves a concrete example of this behaviour. During a 2012 discussion about limit orders, a participant proposed observing the Mt. Gox price and automatically cancelling an order when the difference grew too large. That suggestion did not assume the two prices would remain equal. It assumed divergence was normal enough to require a rule.
Such discussion formed part of the public reputation record around early exchanges. Users compared mechanics and prices in the same thread where operators announced changes, exposing how much market quality depended on rules that differed by venue.
The comparison was a reference point, not a shared clearing system. Mt. Gox could be the dominant venue while another exchange continued to clear trades at its own prices.
What an Early Quote Meant
An archived last-trade value should be read with its venue, pair and time attached. Without those qualifiers it can look more authoritative than it was. Two correct records can disagree because they describe different order books.
Fragmentation narrowed as liquidity deepened and professional traders kept balances across venues, but the mechanism never vanished. Exchanges still maintain separate books. The early period merely made the frictions large enough to see: local currencies, bank transfers, confirmation waits and sparse orders all sat between a visible difference and a completed arbitrage.