“Above a fixed dollar floor, Congress is notified. Below it, Congress is not. The floor was set in 1981; the only amendment since raised it.” — from the editor’s desk
a $14 rubber-and-steel drive belt off a home lawn tractor, the kind sold at any hardware store
the belt's tread logic descends from the continuous track, invented in 1904 by Holt Manufacturing so a steam tractor would not sink into California's soft delta soil — the wheel built to never quite touch ground
Holt's tread patent suit forces a 1925 merger with rival Best Tractor, birthing Caterpillar Tractor Co. out of a patent settlement, not a war contract
a soil problem, solved by a corporate merger, leaves behind one chassis that farms and armies both keep wanting
LINK: Caterpillar's D9, armor-plated by Israel Military Industries, becomes the IDF's standard combat-engineering vehicle for route-clearing operations in the Gaza Strip
the Arms Export Control Act of 1976, passed by a post-Watergate Congress to rein in unchecked Nixon-era arms sales, instead built the Foreign Military Sales channel the D9 purchase runs through, triggering congressional notice only above a set dollar floor
LINK: cash-flow financing under a 1973 emergency act and, since FY1991, the lump sum disbursed within thirty days — appropriated dollars spent by letters of offer years before any hardware ships, for Israel and at times Egypt
The Foreign Military Financing (Israel) line inside the annual State–Foreign Operations Appropriations bill — the ledger that turns a 2016 memorandum into a recurring, review-proof number Congress re-enters rather than re-examines.
Chapter 1: The Endless Loop
18 September 2026
You are holding a drive belt. Fourteen dollars at the hardware store, a loop of black rubber with cord buried in it, and you are holding it because the one on the lawn tractor snapped halfway across the yard on Saturday and the mower deck went quiet. Look at it for a moment before you put it on. It has no beginning. That is the whole of its cleverness: a loop carries power around a corner by never arriving anywhere, and the pulleys it rides on never touch each other at all. Now — hold that thought, because the same idea, turned upside down and cast in steel, is going to carry you a very long way from the yard, and I already know where it ends, and I am not going to tell you.
The upside-down version starts in mud. Stockton, California, 1883: Benjamin Holt and his brothers open a factory and begin building combined harvesters and steam traction engines for the Delta, the islands of peat west of town where, the Haggin Museum will tell you, any number of lucrative crops would grow and the heavy round wheels of the engines that pulled the machinery would sink. Spongy soil, a boiler on wheels, and the wheels going down. Every farmer in the Delta knew the problem. Holt's answer was to stop trying to make a wheel that would not sink and instead build a machine that laid its own road.
Here is how it is built, because the patent is public and I have read it so you don't have to. United States Patent 874,008, "Traction-engine," filed by Benjamin Holt on the ninth of February 1907 and granted the seventeenth of December. Take a chain of steel links — alternating, the narrow ends of one link passing between the wide ends of the next, pinned through with a pivot and a sleeve — and join enough of them to make an endless loop. Wrap the loop around a toothed sprocket at each end. The sprocket has an odd number of teeth, deliberately, so that the same teeth do not take the load every time round and wear evenly forward and reverse. Bolt a flat transverse shoe to the outside of every link, with a flange that bites the ground. Then set the whole weight of the engine on flanged rollers that run along the inside edge of the links, staggered so no two roll in the same plane. That's the trick. The shoes lie flat on the peat like planks on a bog, the rollers ride across the planks, and the machine picks its road up behind itself and lays it down again in front. Your fourteen-dollar belt, rendered in cast iron, with the machine riding on the inside of the loop instead of the pulley riding on the inside of the belt. Same geometry. Different tenant.
The first practical demonstration, by the American Society of Mechanical Engineers' account, was the twenty-fourth of November 1904, in a peat field on Roberts Island — Thanksgiving Day, the collectors' club adds — with the rear wheels taken off a steam engine and tracks bolted on in their place. The following March the company photographer, Charlie Clements, watched it move and said it crawled like a caterpillar, and Holt, who knew a trademark when one crawled past, kept the word. Patented and in production by December 1907. The ASME designated the machine a historic engineering landmark in 1981; it sits at the Haggin in Stockton, where the smell of machine oil still clings to it.
I should be honest with you about the word "invented," since this serial runs on receipts. The track was invented, as most things are, by several people at once. A firm in Grantham, England, patented a crawler track in that same year of 1904 and fitted it to a War Office tractor by 1907, and the Tank Museum at Bovington, which holds that machine, records a rival story in which soldiers coined the word caterpillar first. There were older American patents too — a Mr. Lombard's — which pre-dated Holt's, a fact that is going to matter a great deal in the next chapter, when somebody buys them for twenty thousand dollars. What Holt owned, and what the museum's wording carefully gives him, was the first commercially successful track-type tractor: the self-laying continuous track that gave more bearing area than a round wheel and kept the engine out of the Delta. He owned the word, the drawings, and the customers. He did not own the idea, and the courts were about to spend thirteen years explaining that to him.
So: a belt in your hand, a loop of steel in a peat field, a patent with an odd number of teeth. A soil problem, solved. What the solution set loose was not a machine but a lawsuit, and a lawsuit, unlike a tractor, does not stop when the money runs out.
Last time I left you in a peat field with a patent. I'd like you to leave the field now and come indoors, to a courtroom, because the company whose name is on the machine that is going to matter later was not born in a field, or in a war, or in a boardroom of visionaries. It was born in a settlement. I find that so satisfying I can barely stand it.
The rival was named Best. Daniel Best built steam traction engines and harvesters a few miles from Holt, and in 1905 the two firms went to law over patents — by the collectors' club's account a suit for infringement, by the trade historians' a fight over a steam-driven auxiliary motor that ran the cutting and threshing works on a combine. The trial ran, the appeal ran, a retrial loomed, and in 1908 everyone did the arithmetic. Daniel Best retired, gave a third of his company to his son, Clarence Leo Best, and sold the other two-thirds to Benjamin Holt. The first war ended with the loser buying the winner. Now — hold that thought, because C. L. Best did not care for working inside a Holt-controlled company, and in 1910 he walked out and started the C. L. Best Gas Traction Co. at Elmhurst, near San Leandro, and built crawlers of his own.
In 1915 he showed one at the state fair. He called it the Tracklayer. Holt sued him for patent infringement, and here is where the older patents from last chapter earn their keep: Best bought the Lombard patents, the ones that pre-dated Holt's crawler claims, for twenty thousand dollars, and countersued. Watch what a patent war does to a patent. Two firms, each holding paper that says the other cannot build the thing both are building, and a courthouse that will not say which paper wins. The upshot, in the club's words, was a settlement: a cash payment to Best, plus a licence authorising Best to use all the Holt patents. Holt had also sued over the word itself, and Best was barred from calling his machine a caterpillar. So by 1918 the patent Holt had fought thirteen years to defend was licensed to the man he'd fought, and the only thing the lawsuits had actually protected was a noun. The litigation, 1907 to 1918, is reliably estimated to have cost the two companies about a million and a half dollars, at a time — the club says this, not I — when the dollar had real value.
Here is how a company like that runs, since you are going to watch two of them stop. Routine operating capital came from lines of credit; you borrow against the season's orders and repay when the harvest sells the machines. That works when the orders are real. During the war the military contracted with Holt for 24,791 tractors, and by the armistice on the eleventh of November 1918 only 9,771 had been built — a factory tooled for a customer who had just stopped calling. Then the government sold its surplus tractors into the civilian market, at surplus prices, against the very firm that had built them. Benjamin Holt died in December 1920. The agricultural depression arrived the same year. Best, with better books and a newer product line, was borrowing too. Two makers, one market, both of them living on credit and excess capacity.
Enter the bond broker. Harry H. Fair, of Pierce, Fair & Company of San Francisco, held a seat on Best's board and had funded Best's debt; then he was engaged by the Holt company and, the club says, quickly became aware of its precarious condition. He proposed that the two consolidate, because it was possible that both would go out of business if they did not. He made his overtures to the Holt side quietly. In April and May of 1925 the two firms did not so much merge as reorganise and combine their financial ownership, transferring everything into a new entity: Caterpillar Tractor Co. C. L. Best became chairman. Holt's people had the worldwide name, the trademark, and the bigger factories; Best's had the money and the dealers. Headquarters at San Leandro, then Peoria by 1930.
So read the route card honestly with me. The patent war did not force the merger. The patent war ended in 1918 with a licence and an empty treasury on both sides, and what forced the merger seven years later was a war contract that never finished, a surplus that undercut the survivors, and a broker who could count. Two things are nonetheless true and I want them side by side: the company was made by lawyers and financiers reconciling two ledgers, not by a general placing an order — and the machine they were reconciling over had already been to a war. That second fact is the one that keeps the chassis alive for the next hundred years, and it deserves a chapter in which nobody sues anybody.
Three chapters in, and it is time we did the bookkeeping. Put the belt down on the workbench — you can put it on the mower later — and lay the receipts beside it, because a chain is only a chain if the links hold when you pull on them from the far end, and I want you to pull.
Receipt one: a soil problem. Peat that swallows a round wheel. Receipt two: a machine that lays its own road, patented in 1907 with an odd number of teeth on the sprocket, and — the record is firm on this — one of several tracks patented in the same few years by several men in two countries. Receipt three: a patent war that protected nothing but a noun, and a merger in 1925 that was a bond broker's rescue of two firms drowning in surplus. Now put those side by side and notice what none of them says. None of them says anything about an army. And yet the surplus that drowned them was the army's surplus. Where did the army come in?
Here is how a gun moves, in 1915, which is the year that matters. Heavy artillery on the Western Front was hauled by horses, and horses die in mud, and the Holt track does not. The National Army Museum in London, which holds a lantern slide of a Holt tractor towing a gun on the Western Front in 1917, states the arithmetic plainly: the Holt Company of California made the tractor models, the 75 and the 120, that the British Army adopted during the war, and by 1916 the British Expeditionary Force had over a thousand of them in use for the purpose of hauling its heavy artillery. The Australian War Memorial holds a photograph of a 75-horsepower Holt hauling guns and stores across the Sinai desert. The collectors' club counts the American side: 1,800 of the Model 45, 1,500 of the Model 75 and 90 of the Model 120 sold to the U.S. Army, more than five thousand tractors in wartime production all told, 2,100 of them to the Allies. The machine that would not sink in a Delta peat field would not sink in Flanders either. Nobody redesigned it. The army simply bought the farm tractor and hitched a howitzer to it.
You will have heard the tidier story — that the Holt tractor is the father of the tank. I like a tidy story as much as the next narrator, which is why I checked it against the people who keep the tanks. The Tank Museum at Bovington, in its own words, says the tank evolved in 1915 through Diplock's Pedrail system and the Bullock Creeping Grip Tractor Company of Chicago — another crawler, another firm — and it says so while dismissing an English claimant, which tells you the museum is not in the business of flattering anyone. So the honest line is narrower and better: the war did not turn Holt's tractor into a weapon. The war bought Holt's tractor as a tractor, by the thousand, to pull the weapons, while an agricultural machinery firm in Lincoln built the weapon on somebody else's track. A tracked chassis, it turns out, is a thing armies want for the same reason farmers want it. It does not care what it carries.
Now the reconciliation, and this is the part I've been saving. Farms and armies did not want two machines. They wanted one. The 24,791 tractors the military contracted for, and the 9,771 actually built by the armistice, were the same product the Delta had been buying since 1908; when the government dumped its surplus into the civilian market after 1918, it was competing with Holt's own catalogue, using Holt's own machines. The overlap that ruined the company is the overlap that made it indispensable. And after the merger the chassis kept doing exactly what it had done: 1931, the Diesel Sixty, fourteen sold in its first production year; then the D-numbered tractors, until in 1954 the company introduced what the club calls its largest tractor to date, the D9, in production from 1955 in two transmissions. A farm machine. A quarry machine. A road-building machine. And, wherever a state has a mud problem, an army machine — the same iron, the same self-laid road, with something bolted on.
What the columns show, laid flat: a soil problem produced a loop of steel; the loop of steel produced a lawsuit; the lawsuit produced a licence and a bankruptcy; the bankruptcy produced a company; and running under all of it, from 1915 onward, is a customer who never needed the machine changed, only delivered. Farms and armies keep wanting the same chassis, and the company keeps selling it to both, and the ledger does not distinguish between a furrow and a trench because a track doesn't either.
I have not told you which state, or which army, or what gets bolted on. The map lights only what we have walked. But you will have noticed that the last thing every chapter has done is add weight to the machine, and that weight, in the modern version, is measured in tons of steel plate.
The D9 is a bulldozer. I want to begin there, flatly, because everything that follows depends on a machine whose catalogue entry is a bulldozer and whose paperwork, as we will see, is not. The Israel Defense Forces describe theirs on their own website, in an article dated the twenty-fifth of February 2025: a modernised Caterpillar D9R, reinforced for safety and durability in combat conditions, operated by the Combat Engineering Corps. More than eight metres long, four and a half wide, about four high. An eighteen-litre diesel making 474 horsepower. Additional armour protection, and a ripper on the back for breaking concrete. Tasks, in the army's own list: neutralising improvised explosive devices, clearing debris after strikes, opening safe passage routes, removing obstacles, demolition. Variants, the army says, have served since the Six Day War.
Here is how the plate gets on, since the machine does not leave Illinois wearing it. Human Rights Watch, in a letter to Caterpillar dated the twenty-eighth of October 2004, set out the route: the IDF purchases D9s through the U.S. government's Foreign Military Sales program, and they are then armoured by Israel Military Industries in Israel. So the sequence is three signatures. A government-to-government sale, administered by the United States. A delivery of a commercial bulldozer. A state-owned Israeli company that bolts the armour, the glass, and the cab protection on afterward. Caterpillar's own line, as HRW reported it, was that the company has neither the legal right nor the ability to monitor and police individual use of its equipment, and that comment on political conflict in the region is best left to governmental leaders. Hold that thought — not the sentiment, the structure. The maker sells a tractor. The buyer is a government. The armour is someone else's. Nobody in the chain is holding a weapon, and a weapon is what comes out the other end.
What the machine does is a matter of record, and I'll let the institutions carry it. Human Rights Watch's report of the seventeenth of October 2004, Razing Rafah, counted over 2,500 Palestinian houses demolished in the Gaza Strip since 2000, nearly two-thirds of them in Rafah on the Egyptian border, 298 of them in May 2004 alone. Armoured D9s razed blocks of homes at the edge of the camp, incrementally widening a buffer zone that was then up to three hundred metres deep, and the IDF, HRW found, destroyed fifty-one percent of Rafah's roads, usually by dragging the ripper from the back of the D9. The organisation's conclusion, in its words: in most of the cases, the destruction was carried out in the absence of military necessity. Twenty years on, in a 154-page report of the fourteenth of November 2024, the same organisation described deliberate, controlled demolitions of homes and civilian infrastructure with the apparent aim of creating an extended buffer zone along Gaza's perimeter and a corridor bifurcating the Strip. The army calls the task opening routes. The report calls it demolition. Both are describing the blade.
🚬
“The official line is straightforward and I'll give it fairly: the D9 is a commercial bulldozer sold to a designated major non-NATO ally through a lawful program that Congress created, Caterpillar adds nothing military to it, the armour is an Israeli decision made in Israel, and the machine clears explosives that would otherwise kill soldiers and civilians alike. Everything in that sentence is true.”
Everything in that sentence is true, and the fold answers it anyway, because the question was never whether anyone did anything unlawful. It was who signed. In February 2025, according to the Congressional Research Service, the Administration invoked Section 36(b) of the Arms Export Control Act to declare that an emergency existed requiring the sale to Israel of, among other things, Caterpillar D9 bulldozers — four Foreign Military Sales cases at once, totalling nearly four billion dollars, of which the D9 case, per the Defense Security Cooperation Agency's own release of the twenty-eighth of February, was estimated at $295 million. CRS also records that the ranking members of the foreign affairs committees had placed a hold on at least two of the four cases, including the bulldozers, and that the emergency declaration proceeded past it. So the plate is bolted on in Israel, yes. But the bulldozer underneath it is sold by the United States government, under a United States statute, on a United States form, with a United States clock attached — and the clock, on this occasion, was declared not to be running.
That statute is the door. It was written by a Congress that intended, on paper, to slow exactly this kind of sale down, and what it built instead is the channel the sale runs through. It has a floor in it, a dollar amount below which nobody is told anything at all, and the floor has a birthday.
Every clock has to be wound by somebody, and the clock on an arms sale was wound, first, in December 1974 — the winter after a presidential resignation, by a Congress in the mood to attach conditions to things. The Foreign Assistance Act of 1974 gave the legislature a look at a government arms sale before it closed and a way to stop it. The floors, by the time the U.S. Code's own amendment notes pick them up, stood at $25 million for a sale and $7 million for major defense equipment. Below those numbers, nothing. Above them, the executive had to tell Congress and wait.
Then came the fight over the bigger bill, and I want you to notice who won it. In 1976 Congress sent President Ford a sweeping rewrite, S. 2662, and on the seventh of May he sent it back. His veto message is in the Senate's files: the bill, he wrote, would seriously obstruct the exercise of the President's constitutional responsibilities for the conduct of foreign affairs; it imposed an arbitrary arms sale ceiling and it obstructed U.S. industry from competing fairly with foreign suppliers. Congress did not override. It rewrote. The replacement, H.R. 13680, dropped the ceiling and most of the provisions the President had objected to, kept the legislative veto over major government sales, and he signed it on the thirtieth of June 1976 as Public Law 94-329, the International Security Assistance and Arms Export Control Act. His signing statement the next day contains the sentence I would frame if I framed things: although I am accepting this provision, I reserve my position on its constitutionality if the provision should ever become operative. That is the birth certificate of the Arms Export Control Act. A leash, accepted with a note reserving the right to slip it.
Here is how the clock runs today, from the statute, because the statute is public and shorter than you'd fear. Section 36(b), Title 22, section 2776 of the U.S. Code. Before the government issues a Letter of Offer for a Foreign Military Sale of defense articles or services worth $50 million or more, or major defense equipment worth $14 million or more, or design and construction services worth $200 million or more, the President transmits a certification to Congress — the buyer, the item, the price. The letter of offer may not be issued for thirty calendar days. For NATO members, Japan, Australia, South Korea, Israel and New Zealand the floors are higher — $25 million, $100 million, $300 million — and the wait is fifteen days. To stop the sale Congress must, within that window, enact a joint resolution prohibiting it. Enact: pass both houses and obtain the signature of the President who proposed the sale, or override his veto by two-thirds. If Congress does nothing, the sale proceeds. And if the President determines that an emergency exists requiring the sale immediately in the national security interests of the United States, and writes down a justification, the waiting period is waived and the letter goes out. Before any of that, by a practice that began with a Defense Department letter in 1976 and sits in no statute, the State Department gives the two foreign affairs committees an informal look, twenty to forty days ahead, in which staff may ask questions. That is the entire mechanism. A floor, a clock, a resolution that needs the seller's own signature, and a switch marked emergency.
Now the birthday, since the masthead promised one. The $14 million and $50 million floors were set in 1981 by Public Law 97-113, doubling the 1976 figures. The $200 million design-and-construction floor was added in 1980. The higher floors and the fifteen-day clock for NATO, Japan, Australia and New Zealand were added in 2002; South Korea joined that list in 2008 and Israel in 2010. In 1986, after the Supreme Court had struck down the concurrent-resolution veto in another statute, Congress substituted the joint resolution — the version that needs a signature. The floors themselves have not moved since 1981. A dollar figure fixed in the first year of the Reagan administration has, in forty-five years, needed no updating, and I will leave you to decide whether that is because the machinery was perfectly calibrated at birth.
🚬
“The official line: Congress designed every inch of this, in public, and has amended it again and again; every case above the floor is certified in writing; the committees get an informal look before the formal one; and Congress retains, at every moment, the power to legislate a sale out of existence. The executive is not evading oversight. It is complying with the oversight Congress wrote.”
The fold agrees with him, which is the trouble. The Congressional Research Service, in its standing report on the process, states that Congress has never successfully blocked a proposed arms sale via a joint resolution of disapproval. Not once, in fifty years, under either party's presidents. The instrument built to stop a sale requires the sale's proponent to sign the stop order. What the 1976 law did instead — and this is not a criticism, it is a description — was build the channel: a Foreign Military Sales case now has a form, a floor, a clock and a filing cabinet, and a bulldozer sold on that form has been reviewed, in the precise sense that a certification was transmitted and a number of days elapsed. Below the floor the clock never starts. Above it the clock runs out. In an emergency it is waived. There is no fourth setting.
Which leaves one question the statute does not answer: with what money? A Foreign Military Sale is paid for by the buyer, and for one buyer the money arrives before the fiscal year does, in a lump, and earns interest while it waits.
A confession before we begin, because this serial runs on receipts and the receipt I went looking for does not exist. I was sent to find a side letter from 1988. There is no side letter from 1988. What there is, in the Congressional Research Service's standing report on aid to Israel, is older and stranger and in two parts, and I'd rather give you the true machinery than the tidy date. The two parts are these: the money arrives early, and the money is spent before it arrives. Neither was set in 1988. One was set in 1973 and the other in 1990, and between them they explain why the bulldozer's invoice is, in a sense I will make precise, already paid by a Congress that hasn't met yet.
Part one is called cash flow financing, and here is how it books a dollar. Section 23 of the Arms Export Control Act lets the President finance foreign purchases of defense articles on such terms and conditions as he may determine. Ordinarily a country signing a Letter of Offer and Acceptance for a multi-year purchase must set aside the full price up front from that year's grant. Under cash flow financing it sets aside only the installment due this year. The rest of the contract is a claim on grants Congress has not yet appropriated. The Government Accountability Office, examining the same arrangement for Egypt in 2006, put it in a sentence I admire for its plainness: the arrangement allows the United States to enter into contracts in advance of — and in excess of — current FMF appropriations. For Egypt, as of March 2006, the letters of offer anticipating future funding totalled about two billion dollars, some not due for full payment until 2011. That is what the mechanism is: a signed contract today, an unwritten appropriation tomorrow, and a future Congress that arrives to find the bill already on the desk.
Where it came from is in the CRS's own text box. After the 1973 war Congress appropriated $2.2 billion for Israel in Public Law 93-199, the Emergency Security Assistance Act of 1973, whose Section 3 provided that military sales credits to Israel be extended on such terms and conditions as the President may determine and without regard to the Foreign Military Sales Act — which at the time capped credit at $250 million a year to any one recipient. President Nixon, in two determinations in April and July 1974, allocated $1.5 billion of it as grant and $700 million as loan. Eighteen months later, CRS records the Ford Administration's agreement, in the words of the New York Times, that the cost of new equipment would be met through the aid just approved by Congress as well as the aid that will be approved by future Congresses. There is your instrument. Not a letter; a determination, under a 1973 emergency act, that the ordinary cap need not apply. Egypt received the same benefit from President Carter in 1979, and as of 2015 CRS could say that only Israel and Egypt were authorised to use it. The Obama Administration announced Egypt's would end from fiscal 2018; the February 2026 CRS report on Egypt allows that it may since have been reinstated. So "Israel alone" is not what the record says. What the record says is: two countries, one of them continuously since 1974.
Part two is the lump sum. CRS again: since fiscal year 1991, under Public Law 101-513, Congress has mandated that Israel receive its Foreign Military Financing in a lump sum during the first month of the fiscal year; the appropriations language reads that the funds shall be disbursed within 30 days of enactment. Once disbursed, the aid is transferred to an interest-bearing account with the U.S. Federal Reserve Bank. The earlier CRS issue brief on the subject, from 2005, gives the origin and the cost: the appropriation bill signed on the fifth of November 1990 provided the lump sum, which allows Israel to invest the funds in U.S. Treasury notes and earn interest; AID officials estimated at the time that borrowing to make the early payment cost the United States between $50 million and $60 million a year; Israel was reported to have earned about $86 million in Treasury note interest in 1991. Israel used the interest, CRS says, to pay down bilateral debt to U.S. agencies; as of September 2024 it no longer had any outstanding sovereign debt to the United States, and the interest cannot be used for procurement inside Israel. Put the two parts together and watch the dollar move. It is appropriated in February; it is disbursed in full by March; it sits at the Federal Reserve earning interest; and the contract it will eventually pay was signed years ago against it, by a letter of offer that treated it as already there.
🚬
“The official line: every word of this is lawful and public. Section 23 authorises it, the appropriations act mandates the thirty days in plain text every year, and the purpose is not concealment but planning — a partner buying aircraft on a decade's schedule needs to sign a decade's contract. Congress can strike the clause any February it likes.”
It can, and the fold explains why it doesn't. CRS's most recent report, weighing what happens when the current agreement ends, notes that with tens of billions of dollars in sales already agreed since October 2023, and with Israel's ability under U.S. law to use cash flow financing to stretch payments on those contracts over years, a new memorandum may be necessary to finance existing payment schedules on equipment already purchased. Read that once more slowly. The contracts have been signed. The payments are scheduled. The appropriation that will meet them is a future Congress's, and the future Congress's choice is not whether to buy the bulldozer but whether to default on it. That is what cash flow financing does to a legislature's discretion: it converts it, one signed letter of offer at a time, into a payment schedule. The instrument was not set once in 1988. It was set in 1973 and has been re-signed every fiscal year since 1991, thirty days at a time.
Which brings us to the line itself — the actual sentence, in the actual bill, that Congress re-enters each year. It is one paragraph long. It has a number in it that was fixed in 2016 and has not changed since, and you have already paid for the belt.
Here it is. Public Law 119-75, enacted the third of February 2026, a six-bill package of which Division F is the National Security, Department of State, and Related Programs Appropriations Act, 2026. Under the heading Foreign Military Financing Program, subsection (c), one word in capitals — ISRAEL — and then the sentence: of the funds appropriated by this Act under the heading "Foreign Military Financing Program", not less than $3,300,000,000 shall be available for grants only for Israel: Provided, That funds appropriated by this Act under the heading "Foreign Military Financing Program" and made available for assistance for Israel shall be disbursed within 30 days of the date of enactment of this Act: Provided further, that to the extent the Government of Israel requests, the grants shall be available for advanced weapons systems, of which not less than $250,300,000 shall be available for the procurement in Israel of defense articles and defense services. That is the destination. A filing cabinet, as promised. One paragraph, re-entered every year, with a number in it.
Here is where the number comes from, and how the paragraph is assembled, because a line in an appropriations bill is built the way a track is built: links, pinned, in an endless loop. On the fourteenth of September 2016 the United States and Israel signed a ten-year memorandum of understanding. The White House fact sheet of that day gives the terms: $38 billion over fiscal years 2019 through 2028 — $33 billion in Foreign Military Financing and $5 billion in missile defense — disbursed in equal increments of $3.3 billion in FMF and $500 million in missile defense funding each year, replacing a $30 billion memorandum signed in 2007. The fact sheet also records the decision to discontinue two anomalies in the relationship, Off-Shore Procurement among them: the portion of American grant money Israel had been permitted to spend in Israel. Two years later, on the first of October 2018, the State Department announced the ten-year period had begun and that the United States will set funding for Israel at levels of $3.3 billion in Foreign Military Financing over each of the next ten years. Note the verb. Will set. A memorandum is not a law. It is an executive promise to ask, and a promise that Congress will be asked, every year, for the same number. Fiscal 2024: $3.3 billion, with $725.3 million for procurement in Israel. Fiscal 2025: $3.3 billion, carried at the prior year's level by a full-year continuing resolution, with $450.3 million. Fiscal 2026: $3.3 billion, with $250.3 million. The only figure that moves is the one the memorandum said would shrink, and it shrinks on the memorandum's schedule. Congress re-enters the line. It does not, on the record, re-examine it.
Now put the whole chain on the bench beside the belt, because you've earned it. A peat field that swallowed a wheel. A loop of steel with an odd number of sprocket teeth, patented in 1907 by one of several men who had the idea. A thirteen-year patent war that protected a noun, and a 1925 rescue by a bond broker that produced a company. An army that bought the farm tractor by the thousand to pull its guns, and a chassis that has never since needed to know whether it is in a furrow or a street. A D9, sold as a bulldozer through a government-to-government program under a 1976 statute with a floor in it, armoured afterward by a state company, and described by one army as opening routes and by one human rights organisation as demolishing homes in the absence of military necessity. A clock that has never once, in fifty years, been used to stop a sale. Money that is appropriated in February, disbursed in full within thirty days, held at the Federal Reserve at interest, and spent years earlier by a letter of offer that treated it as already there. And this paragraph, which supplies the money, at a number fixed in 2016.
🚬
“The official line, one last time: this is the most transparent line in the federal budget. The number is public, the memorandum is public, the vote is public, the thirty days are in the text, and Congress has passed it, year after year, at the number the memorandum names. Nothing here is hidden.”
Nothing here is hidden. That is the finding. The Congressional Research Service, describing the machinery of review, states that Congress has never successfully blocked a proposed arms sale via a joint resolution of disapproval; the same service, describing the money, states that with contracts already signed and cash flow financing stretching the payments over years, a new memorandum may be necessary simply to finance existing payment schedules on equipment already purchased. Read together they say what a filing cabinet says: the decision has been taken, it was taken in the form of a schedule, and each year's line is the schedule arriving on time. The floor for telling Congress about a sale was set in 1981 and has not moved. The number in the paragraph was set in 2016 and has not moved. The thirty days were set in 1990 and have not moved. The floor never falls because nothing is placed on it that could make it.
You bought a fourteen-dollar belt on Saturday. It is the same loop, rubber for steel, pulley for sprocket; the track was invented so that a machine would never quite touch the ground, and the ledger was written so that a decision would never quite touch a vote. The belt on your workbench will drive a mower deck. The D9's track will carry eight metres of iron and plate and a 474-horsepower diesel down a street in Gaza, and the $3.3 billion that keeps it in parts was disbursed within thirty days of the third of February, and is earning interest in an account at the Federal Reserve Bank as you read this, and the next memorandum is, in the Congressional Research Service's estimate, a matter of financing what has already been signed.