An Antminer S21 XP, the air-cooled Bitcoin machine Bitmain released in June 2024, draws about 3.6 kilowatts, roughly what two fan heaters pull when left on full. On 7 November 2026, at that day’s network figures, it would have earned about $8.75 of bitcoin in 24 hours. In the same 24 hours it burns 87.5 kWh of electricity: close to $15.90 at the average US household price, £23 at the UK price cap and €17.50 on France’s regulated tariff.
That gap is the honest answer to the question most people bring to a page about bitcoin blockchain and earns a reward, which makes rewriting history very expensive.">mining. Mining works as an industry, and the network has rarely had as much computing power pointed at it, but a household paying retail electricity in the United States, the United Kingdom, France or Italy pays more for the power than the machine brings back. The sum has few parts, and you can redo it yourself once you know the machines, the pools that share out the winnings and the daily “wage” miners call hashprice. Then come the noise and heat at home, the tax rules and the scams that sell mining to people who will never see a machine.
What bitcoin mining is: a lottery that pays in new coins
The previous lesson, proof of work in depth, covered the mechanics, so here is only the short version. A miner collects transactions waiting in its mempool, builds a candidate block and hashes the block header again and again, changing a few bytes each time, until one result falls below the network’s target. Every 2,016 blocks the difficulty adjusts so that a block arrives about every ten minutes on average. As of 7 November 2026 the difficulty stood at 148.36 trillion, after a rise of 5.54 percent at the 29 October adjustment, according to mempool.space.
The winner writes the first transaction of its block, the coinbase, and pays itself two things: the block subsidy, which is newly created bitcoin, and every fee attached to the transactions it included. That is the whole business model of proof of work. Satoshi Nakamoto chose the mining image on purpose: section 6 of the white paper calls the steady issuance of coins “analogous to gold miners expending resources to add gold to circulation” (our walk through the Bitcoin white paper summarises the rest).
Everything below takes those rules as given. The question here is economic: what it costs to buy a ticket in that lottery, and who can still afford to play.
Where a miner’s money comes from: 3.125 BTC per block and a sliver of fees
The subsidy has been 3.125 BTC since the April 2024 halving
Block 840,000 was mined on 20 April 2024 at 00:09:27 UTC by the ViaBTC smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">pool. It was the fourth halving of bitcoin: from that block on, the subsidy fell from 6.25 to 3.125 BTC. That particular block paid 40.75 BTC in total, of which 37.63 BTC were fees, an exceptional amount, per mempool.space block data. It did not last: within about a week, fees went back to a small fraction of a bitcoin per block, and every machine on the network was earning about half of what it had earned the week before, unless the price of bitcoin rose to compensate.
The subsidy halves every 210,000 blocks, so the next cut is due at block 1,050,000. At ten minutes per block that lands around 2028; the exact date depends on how fast blocks arrive between now and then. A later lesson on the halving schedule will go through the history and the supply curve in detail.
Fees made up about 0.6 percent of revenue in early November 2026
Over the 1,008 blocks to 7 November 2026, miners collected about 3,170 BTC in total rewards, of which only about 20 BTC came from fees, according to mempool.space’s reward statistics. That is about 0.02 BTC per block, or 0.6 percent of revenue. Hashrate Index counted a similar share, 0.55 percent, for the week covered by its roundup of 2 November 2026 (about 18 BTC of fees against roughly 3,268 BTC of block rewards).
So in late 2026, miners live almost entirely off the subsidy. How users set fees and why they spike during busy periods is the subject of our lesson on the mempool and transaction fees; for a miner, the point is simpler. A quiet mempool means the halving schedule, not user demand, decides most of the income.
ASIC vs GPU: why a graphics card no longer mines bitcoin
From home computers to single-purpose chips in four years
In 2009 bitcoin was mined on ordinary processors, with the same software that held the wallet. Within about a year, hobbyists found that graphics cards, built to run thousands of small identical calculations in parallel, were far better at repeating one hash function. Programmable FPGA boards followed in 2011. Then, in early 2013, the first commercial ASIC miners reached customers: the Avalon machines built by the team that went on to found Canaan were among the very first, delivered in January 2013 after a project started the previous autumn.
An ASIC, or application-specific integrated circuit, is a chip designed for one job. A Bitcoin ASIC computes SHA-256d, the double SHA-256 that Bitcoin’s proof of work uses, and nothing else: it cannot run a spreadsheet, render a game or mine a coin built on another algorithm. In exchange it hashes with an efficiency no general-purpose chip comes close to. Point a modern gaming card at the Bitcoin network in 2026 and its share of the total hashrate rounds to zero; it would need to run for far longer than a lifetime to expect one block.
What J/TH means, and the machines on sale in 2026
One number decides whether a machine survives: its efficiency, written in joules per terahash (J/TH). It is simply the power draw divided by the hashrate. An S21 XP rated at 270 terahashes per second (TH/s) for 3,645 watts works out at 3,645 / 270 = 13.5 J/TH. Lower is better: fewer joules of electricity for the same amount of work, so a lower bill for the same share of the lottery.
| Model (Bitmain) | Hashrate | Power | Efficiency | Cooling | Release date (Hashrate Index) |
|---|---|---|---|---|---|
| Antminer S21 XP | 270 TH/s | 3,645 W | 13.5 J/TH | air | 17 June 2024 |
| Antminer S21 XP Hyd. | 473 TH/s | about 5,680 W (derived) | 12.0 J/TH | water | n/a |
| Antminer S23 | 318 TH/s | 3,498 W | 11.0 J/TH | air | 27 May 2025 |
| Antminer S23 Hyd. | 580 TH/s | 5,510 W | 9.5 J/TH | water | 27 May 2025 |
These are the manufacturer’s listed specifications as compiled by Hashrate Index’s hardware overview and its rig database (checked 7 November 2026), not lab measurements; Bitmain’s own shop pages could not be read for this guide. Hashrate Index gives the S21 XP’s tolerances as about 3 percent on hashrate and 5 percent on power, and the S21 XP Hyd. power figure above is simply 473 TH/s multiplied by 12 J/TH. Bitmain is not alone: MicroBT (Whatsminer) and Canaan (Avalon) sell competing machines, which this table does not compare.
Water-cooled “Hyd.” models reach the best efficiency but need a water loop and an outdoor dry cooler. They are built for containers and data halls, not spare bedrooms.
Where the GPUs went: other coins, then the Merge
Once ASICs took over Bitcoin, graphics cards moved to coins whose algorithms were harder to put into a dedicated chip. By far the largest outlet was Ethereum, until 15 September 2022, when the Merge switched it from proof of work to proof of stake. In ethereum.org’s words, “Mining is no longer the means of producing valid blocks”, and the network’s energy use fell by about 99.95 percent. Overnight, a vast fleet of GPUs lost its main job. Some still mine smaller coins, and Monero’s RandomX algorithm was designed to favour ordinary processors, as the previous lesson explains, but none of that touches Bitcoin.
Machines age fast
Each new generation cuts the joules per terahash, and the older machines are the first to lose money when revenue falls. Hashrate Index measures this with an “energy hashprice”: what one megawatt-hour of electricity earns when it is fed into machines of a given efficiency. In its roundup of 2 November 2026, a megawatt-hour earned about $116 in fleets below 14 J/TH, $84 at 14 to 19 J/TH, $63 at 19 to 25 J/TH and only $43 at 25 to 38 J/TH.
The same power bill buys almost three times more revenue in a 2024-2025 machine than in a model from a few generations earlier. That is why a second-hand miner that looks cheap often is not: its price reflects the electricity it will waste. A miner’s hardware is closer to a delivery van than to gold. It wears out, and newer models make it obsolete.
Bitcoin mining pools: how miners share the luck
Solo mining is a lottery ticket
On 7 November 2026 the network was running at about 1,014 exahashes per second (EH/s), or just over one zettahash, according to mempool.space. One S21 XP at 270 TH/s is about 0.0000266 percent of that. With 144 blocks a day on average, simple arithmetic gives that machine one block every 26,000 days or so, about 71 years. That is an average, not a schedule: the block could come tomorrow or never.
No business that pays its electricity every month can live with that variance. Some hobbyists do run small open-source boards of a few TH/s connected to solo pools, knowing it is a ticket with very long odds rather than an income. Everyone who mines for money joins a pool.
Shares: proving your work to the pool
A pool hands each miner a block template and a much easier target than the network’s. Every hash that beats the easy target is a “share”: it proves the machine is doing real work on the pool’s block, even though it is not good enough to win anything. The pool counts shares for each customer. Now and then one of those hashes also beats the real network target, the pool wins the block, and the reward flows into the pool’s accounts. How it then reaches the miners depends on the payout scheme.
FPPS, PPS, PPLNS: who carries the bad luck
| Scheme | What you are paid for | Who carries the luck | Example |
|---|---|---|---|
| PPS (pay per share) | A fixed amount per valid share, paid from the pool’s balance whether or not it finds blocks | The pool | Classic PPS often left transaction fees out |
| FPPS (full pay per share) | Per share, the subsidy plus an average of transaction fees | The pool | Braiins Pool, 2.5 percent fee |
| PPLNS (pay per last N shares) | A cut of each block the pool actually finds, in proportion to your shares in the last window of N shares | The miner: income follows the pool’s luck | OCEAN uses a PPLNS-type scheme it calls TIDES |
| Score-based (historical) | Shares weighted by when they were submitted in a round | The miner | Slush’s pool, to discourage “pool hopping” |
Under PPS, the Bitcoin wiki notes, the scheme ends up transferring all of the risk to the pool operator: the pool pays for every share and absorbs the weeks without a block. That service is why pools charge a fee. Braiins Pool, for instance, pays FPPS with a 2.5 percent fee, settles rewards per UTC day as if the network had found its average of 144 blocks, and states on its rewards page that “Proceeds from transaction (network) fees are shared with miners.”
PPLNS works the other way round. Fees are often lower, but a pool that hits a dry spell pays less, and one that gets lucky pays more. For a small miner the choice is mostly about how much variance the budget can take, plus the minimum payout threshold, the fee and whether the pool pays in bitcoin on-chain or by another route. No scheme changes the expected value much; they change who absorbs the swings.
Who mines today: three pools find about 60 percent of blocks
| Pool | Blocks, last week (957) | Share | Blocks, last month (4,593) | Share |
|---|---|---|---|---|
| Foundry USA | 221 | 23.1% | 1,172 | 25.5% |
| AntPool | 189 | 19.7% | 940 | 20.5% |
| F2Pool | 165 | 17.2% | 774 | 16.9% |
| ViaBTC | 99 | 10.3% | 432 | 9.4% |
| SpiderPool | 57 | 6.0% | 252 | 5.5% |
| OCEAN | 52 | 5.4% | 163 | 3.5% |
| MARA Pool | 50 | 5.2% | 239 | 5.2% |
| Luxor | 44 | 4.6% | 208 | 4.5% |
| SECPOOL | 38 | 4.0% | 189 | 4.1% |
| Braiins Pool | 15 | 1.6% | 72 | 1.6% |
Source: mempool.space mining dashboard, one week and one month to 7 November 2026. The three largest pools found 60.0 percent of blocks over the week and 62.9 percent over the month. These figures count blocks found, identified by the tag each pool writes in its coinbase transaction. They do not show who owns the machines: a pool is a coordinator, and its customers can leave for another pool in minutes.
Those labels may also understate concentration. In a post of September 2024, the Bitcoin developer b10c showed that nine pools, including AntPool, SpiderPool, SECPOOL, Luxor and Braiins, were sending the same or very similar block templates to their miners, with a combined 37.6 percent of hashrate over the month he studied. He treated it as a sign of “proxy pooling”, while stressing that it does not prove a single company controls them all (b10c, template similarity).
Stratum V2: who chooses the transactions
Miners talk to pools through a protocol called Stratum. The first version, from 2012, sends jobs in plain text and lets the pool alone build the block template, which means the pool decides which transactions go in. Stratum V2 is a binary replacement with two main changes. Traffic between miner and pool is encrypted and authenticated, which closes the door to an attacker on the network quietly redirecting someone’s hashrate. And an optional Job Declaration feature lets the miner, not the pool, pick the transactions, while the pool still handles the payouts. The project also says it cuts bandwidth substantially for pools and miners.
As of 7 November 2026, the project’s site lists Braiins Pool and DMND as production pools speaking Stratum V2, plus several solo pools such as CKPool and Public Pool. None of the four largest pools in the table above appears on that list. That matters beyond engineering: when a handful of pools pick the transactions for most blocks, they are also the point where pressure to exclude certain transactions would be applied. Concentration as an attack risk is the subject of a later lesson on 51 percent attacks.
Is bitcoin mining still profitable?
At home, with retail electricity in the United States, the United Kingdom, France or Italy, no: as of 7 November 2026, a current machine burns more in power than it earns, before you even pay for it. Mining can still make money for operators that buy electricity at a few cents per kWh and run recent hardware, and even for them the margin in late 2026 is thin.
Hashprice: the miner’s daily wage
Miners sum up their revenue with one figure, the hashprice: how much a unit of hashrate earns per day. Hashrate Index put it at $32.86 per petahash per second per day (PH/s/day) in its roundup of 2 November 2026, down 8.7 percent on the week, or 0.00042635 BTC. The same report quoted Luxor’s forward curve at an average of $33.04 over the following six months, and noted that hashprice was close to break-even for many operators, depending on their power costs and machines.
You can rebuild the figure from public data. With the difficulty of 148.36 trillion and fees averaging 0.02 BTC per block, each terahash per second earned about 42.6 satoshis a day on 7 November 2026 (our calculation from mempool.space data). At the bitcoin price of $76,021 shown by mempool.space at 09:40 UTC that day, that is $32.42 per PH/s per day, in line with Hashrate Index. Hashprice moves with three things: the price of bitcoin, the difficulty and the fees.
Every “bitcoin mining calculator” on the web does the same sum with four inputs. Revenue per day is hashprice times your hashrate. Cost per day is your machine’s power in kilowatts, times 24 hours, times your price per kWh. Then take off the pool fee. If you know those four numbers, you do not need anyone’s calculator, and you can check any seller’s promise against them.
A worked example: one machine, four countries
This is an example, not a forecast and not advice: one Antminer S21 XP at its listed 270 TH/s and 3,645 W, running 24 hours a day, at the network figures of 7 November 2026 (hashprice $32.42 per PH/s per day; bitcoin at $76,021, €68,260 or £56,913 at 09:40 UTC per mempool.space). Daily revenue: 0.00011514 BTC, or $8.75, €7.86 or £6.55. Daily consumption: 87.5 kWh. No machine cost, pool fee, cooling, standing charge or tax is included.
| Country | Household electricity price used | Power cost per day | Revenue per day | Result per day | Result per 30 days |
|---|---|---|---|---|---|
| United States | 18.16 cents/kWh, average residential, August 2026 (EIA) | $15.89 | $8.75 | -$7.14 | about -$214 |
| United Kingdom | 26.32 p/kWh, Ofgem price cap, 1 October to 31 December 2026, direct debit | £23.02 | £6.55 | -£16.47 | about -£494 |
| France | €0.2001/kWh incl. tax, EDF Tarif Bleu Base, 6 kVA, from 1 August 2026 | €17.50 | €7.86 | -€9.64 | about -€289 |
| Italy | 43.43 c/kWh, ARERA reference price for vulnerable customers (Maggior Tutela), from 1 October 2026 | €37.99 | €7.86 | -€30.13 | about -€904 |
Sources for the prices: EIA Electric Power Monthly, table 5.6.A; Ofgem’s cap for October to December 2026 (GB average for direct debit, without VAT, which the government removed from electricity bills for that period; the daily standing charge of 54.83p is left out); EDF’s Tarif Bleu page; ARERA’s release of 29 September 2026. The Italian figure is the regulated reference for vulnerable households, up 37.3 percent from the previous quarter; free-market offers in Italy vary widely, so treat that row as one reference point, not the national average.
In every row the electricity costs more than the bitcoin the machine produces. And the example is generous. It leaves out the machine itself, which costs thousands of dollars, the 2.5 percent or so a pool keeps, the extra power for ventilation, and taxes on whatever is earned. With those included, the hardware is never paid back at these prices. A higher bitcoin price would narrow the gap; a lower price, a rising difficulty or a new halving would widen it. None of this is a forecast, and none of it is advice to buy or not buy anything.
Who still makes money: power at a few cents per kWh
The same sum gives a break-even electricity price. Before pool fees and hardware cost, the S21 XP covers its power at about 10.0 US cents per kWh (about 9 euro cents) at the hashprice of 7 November 2026. The air-cooled S23 breaks even at about 12.3 cents and the water-cooled S23 Hyd. at about 14.2 cents (our calculation from listed specifications). Most household tariffs in the four countries above sit well over those levels.
Industrial miners live below them. They sign long-term power contracts, set up next to stranded or surplus generation, and in some markets are paid to switch off when the grid is short of power. That is why the hashrate in the pool table is dominated by companies with data centres, not by individuals. Home solar does not change the logic as much as it seems: power you could have sold back to the grid or used in your home has a value, and that value is the real cost of feeding a miner.
Noise, heat and power at home
A machine drawing 3,645 watts around the clock is, in physical terms, a 3.6 kW electric heater: almost every watt that goes in comes out as heat. Air-cooled ASICs push that heat out with high-speed fans, and they are loud enough that owners usually keep them in a garage, a shed or a soundproofed box rather than a living room. Water-cooled models are quieter at the machine but need a pump, a radiator and somewhere to dump the heat.
The power draw is the bigger issue. At 120 volts, 3,645 watts means more than 30 amps, well beyond what a standard 15- or 20-amp household circuit in the United States is built to carry; these machines expect a supply of 200 volts or more, not an ordinary 120-volt outlet. Before plugging anything in, a qualified electrician should check the wiring, and a tenant should check the lease. The home insurer and local noise rules matter too. Some owners use the heat in winter, which is a fair use of power they were going to burn anyway, but the bitcoin earned still does not cover the bill at retail prices.
Energy: 138 to 165 TWh a year by Cambridge’s estimates
Two dated figures from the University of Cambridge frame the question. The Cambridge Bitcoin Electricity Consumption Index (CBECI) showed on 7 November 2026 a best-guess estimate of about 165 TWh a year, but the index’s data then ended in early May 2026, so the figure describes the network of last spring, when hashrate was lower. Earlier, for a report published on 28 April 2025, the Cambridge Centre for Alternative Finance surveyed 49 mining companies representing about 48 percent of global mining activity. It put Bitcoin’s electricity use at 138 TWh a year, about 0.5 percent of world consumption, with 52.4 percent of the energy reported by those firms coming from sustainable sources, nuclear included. The arguments about whether that energy is well spent belong to a later lesson on the Bitcoin energy debate.
Cloud mining and “mining apps”: how the scams work
Cloud mining sells you a contract instead of a machine: you pay up front for a slice of hashrate in someone else’s data centre and watch your earnings tick up on a dashboard. Hosting deals in which a company runs a customer’s own, identifiable machines do exist. The problem is that a buyer of “hashrate” usually cannot check that the machines exist, that they are switched on or that the numbers on the screen come from a real pool. That makes the model a natural cover for fake investment platforms.
Mining apps for phones are simpler still. A phone cannot mine bitcoin in any meaningful way: its processor would contribute a vanishing fraction of the hashrate of one ASIC, which itself expects a block every few decades. An app showing bitcoin piling up on your phone is showing a number in its own database. When such an app asks for a fee to “unlock” a withdrawal, the fee is the product.
The best-known collapse is HashFlare, an Estonian cloud-mining service whose sales topped $577 million between 2015 and 2019. According to the US Justice Department, it relied on fake online dashboards and lacked the computing capacity to mine the vast majority of the cryptocurrency it claimed to generate. Its two founders, Sergei Potapenko and Ivan Turõgin, pleaded guilty to conspiracy to commit wire fraud in February 2025. On 12 August 2025, US District Judge Robert S. Lasnik in Seattle sentenced each to 16 months in prison, time they had already spent in custody, plus a $25,000 fine and 360 hours of community service, and the sentences included the forfeiture of assets valued at over $450 million (Justice Department release). Prosecutors had asked for ten years; in late August 2025 they appealed the sentences to the Ninth Circuit, Cointelegraph reported.
The SEC has kept bringing cases. In December 2025 it sued Danh C. Vo, founder and chief executive of VBit Technologies, alleging he raised more than $95.6 million from about 6,400 investors through “Hosting Agreements” for far more mining rigs than VBit operated, and misappropriated about $48.5 million of investor money, according to SEC Litigation Release 26448. In July 2026 it filed partially settled charges against Mining Automatic and its founder Zan Shaikh, alleging about $22 million was raised from more than 380 investors with promised monthly mining returns, while only about 13 percent of the money went to mining expenses, as reported by The Block. These are the SEC’s allegations, not court findings. Paying old investors with new investors’ money is the mark of the Ponzi schemes described in our dedicated guide.
Europe sees the same pitch. On 21 January 2026, Italy’s market regulator Consob suspended for 90 days, as a precaution, the offer to Italian residents of “mining or investment contract plans” by Fyenergy Cryptocurrency Investment Ltd through three websites (Consob resolution 23832).
There is a legal reason these cases go through securities law. In a staff statement of 20 March 2025, the SEC’s Division of Corporation Finance said it did not view solo or pool mining as securities offerings. That statement covers people who mine with their own machines; a contract in which you hand money to a company and wait for it to produce returns is a different arrangement, and the SEC’s complaint over VBit’s hosting agreements treats them as securities.
The warning signs repeat from one case to the next:
- A guaranteed daily or monthly return. Mining revenue moves every day with the price, the difficulty and the fees, so nobody honest can promise a fixed figure.
- A contract priced above what its hashrate earns. Multiply the TH/s on offer by the current hashprice and compare with the price asked.
- A phone app that claims to mine bitcoin for free, or for a small “booster” payment.
- Referral bonuses that pay more than the mining itself.
- Withdrawal fees, “taxes” or deposits demanded before you can take money out.
Anyone who has lost money to such a scheme should also expect a second wave: people who promise to get it back for a fee are usually part of the same fraud. Our guide to avoiding crypto scams covers the wider family of tricks and how to check an offer before paying.
Taxes on mining income in the US and the UK
In the United States, the IRS answered the mining question in Notice 2014-21 of March 2014. Its answer A-8 says that “when a taxpayer successfully ‘mines’ virtual currency, the fair market value of the virtual currency as of the date of receipt is includible in gross income” (IRS Notice 2014-21). Answer A-9 adds that if mining is a trade or business, carried on outside an employment relationship, the net earnings are self-employment income and owe self-employment tax. The value counted as income then becomes your basis: when you later sell the coins, the difference is a capital gain or loss. Whether your mining is a business or a hobby changes what you can deduct; that is a question for a tax adviser, and our guide to crypto taxes in the United States covers the reporting side.
In the United Kingdom, HMRC’s Cryptoassets Manual (CRYPTO21150) first asks whether the mining amounts to a trade, looking at the degree of activity, organisation, risk and commerciality. If it does, the rewards are trading receipts. If it does not, their sterling value when received “will be taxable as income (miscellaneous income)”, minus appropriate expenses, and the miner “may have to pay Capital Gains Tax when they later dispose of them”. See our guide to crypto taxes in the UK for rates and allowances.
Other countries reach similar results by their own routes. France taxes mining profits as non-commercial income (BNC), according to impots.gouv.fr. In Italy, a 2022 ruling of the Agenzia delle Entrate (answer 508 of 12 October 2022) concerned a company planning to mine: it found the activity outside the scope of VAT and treated the rewards as business revenue valued at the exchange rate of the day received. That ruling is about a company and should not be read as the rule for individuals.
Bitcoin mining in five points
- A miner earns the 3.125 BTC subsidy plus fees for each block it finds; in early November 2026 fees were well under 1 percent of that revenue.
- Bitcoin is mined only with ASICs: graphics cards left for other coins, and lost their largest market at Ethereum’s Merge in September 2022.
- Pools turn the lottery into a wage; FPPS and PPS shift the bad luck to the pool, PPLNS leaves it with the miner, and three pools found about 60 percent of blocks in the week to 7 November 2026.
- Hashprice was about $32 to $33 per PH/s per day in early November 2026, which puts the break-even of a current air-cooled machine around 10 to 12 US cents per kWh before costs, below household tariffs in the US, UK, France and Italy.
- Cloud-mining contracts and phone “mining apps” are the most common way people lose money to mining without ever owning a machine.
Two dates to watch: mempool.space estimated on 7 November 2026 that the next difficulty adjustment, at block 975,744 around 13 November, would cut difficulty by about 4 percent, which lifts hashprice a little; the next halving, at block 1,050,000 around 2028, will cut the subsidy to 1.5625 BTC. Hashrate and difficulty, the halving schedule, the energy debate and running your own Bitcoin node each get a lesson of their own later in this path.
