When Could a Worker Leave a Job in Japan?

In 1842, a woman named Saku was placed in three years of service at an inn near Zenkōji, in what is now Nagano. The innkeeper paid fifteen ryō. The money went to Kinnosuke, the man who had previously employed her. He had taken her on a pilgrimage, become ill during the journey, and incurred expenses he could not meet.^[1]

Saku's future work supplied the payment for that difficulty. Her duties as a meshimori-onna combined service at the inn with sexual service. The contract stated when the three-year term would end. It also specified what should happen if she left before then.^[2]

That second part is the starting point for a history of leaving work. A document could assign a worker to an employer for a period, yet departures could still occur. Who would pursue the person? Would someone else have to repay the money? Could the employer transfer the worker elsewhere?

From the early Tokugawa government in the seventeenth century to Meiji courts and lawmakers in the nineteenth, authorities answered these questions differently. The evidence here concerns contracted personal service, not every occupation in Japan. It reveals why neither the existence of wages nor the existence of a contract is sufficient to tell us whether someone was free to go.

The early Tokugawa government confronted service arrangements that could bind people for long periods. The vocabulary requires care. Hōkō covered service in several settings, while nenki hōkō specified a term. A promise ending after a stated number of years existed alongside other forms of personal dependency.

Historian Rômulo Ehalt places the shogunate's legislation of 1616 and 1619 within that complicated setting. The earlier decree capped term-service contracts at three years and prohibited perpetual sales. The later articles addressed trafficking and contracts exceeding that limit. Yet they did not order the general liberation of everyone already in bondage. In some circumstances, someone unlawfully taken and sold was to be returned to a parent or an original master.^[3]

A restriction on one person's power could therefore preserve another person's claim. Returning an abducted servant to the original household repaired the relationship that the abduction had disturbed. It did not necessarily give the servant the authority to select an entirely different future.

That distinction also explains what a term could accomplish. A stated endpoint supplied something against which continued demands might be measured. An arrangement lasting three years was different from one that made no provision for ending at all. But counting the years did not determine who controlled the work during them, or whose consent had created the obligation.

Nor should the early decrees be read as a complete employment code governing every later contract unchanged. Their importance here is the distinction they were attempting to enforce: a bounded period of service was to be treated differently from a permanent sale. Subsequent contracts still have to be read for the particular obligations they imposed.

This is a more useful beginning than dividing everyone into free wage earners and slaves on the basis of one word. Payment, duration, and control over the person could be arranged in different combinations. A worker might have a dated term and little authority to interrupt it. A rule against sale might coexist with a recognized obligation to serve someone already claiming that person's labor.

The decisive test comes when the relationship ceases to operate as expected. An employer who loses the worker can demand several things: the worker's return, substitute service, or money. Which demand other people recognize helps establish what the original promise actually meant.

A contract from 1772 gives the problem a different scale. Jirōhachi placed his son Masaemon in service with Nakamura Rokuzaemon, an inn and transport-office proprietor at Kashiwabara post station. The agreement ran from the second month to the tenth, an eight-month period excluding winter. A guarantor joined the arrangement.^[4]

The document specified wages and a bolt of summer cotton. It also promised service by day and night according to the employer's wishes. If the employer found Masaemon unsuitable, he could dismiss him; the people guaranteeing the arrangement undertook to provide a replacement or money, according to the employer's preference.

A short term could thus contain a demanding claim over the worker's time. The seasonal boundary answered how long the engagement was intended to last. It did not establish a working day within that period or give the son an equivalent power to dismiss the employer.

The guarantees are equally revealing. This was more than a private promise made by Masaemon in isolation. His father and the guarantor supplied people to whom an unsuccessful employer could turn. Their participation connected the workplace with another household and another source of responsibility.

For the employer, that widened the promise. Losing the worker did not necessarily mean losing every claim arising from the arrangement. Someone remained identified as answerable for a replacement or a payment. For the people placing him, the agreement could create consequences that continued after he ceased to work there.

The document therefore kept two questions together: would the promised service be supplied, and who would bear the loss if it was not? The answer to the second could influence the first. An employer did not have to rely solely on keeping the worker physically present if other people had reasons to ensure that the undertaking continued.

We should also keep the eight months in view. This particular engagement was designed to end without becoming a lifetime relationship. There was nothing inherently contradictory about a specified ending and substantial dependence before it. The historical problem is how much of the interval the worker could control.

That makes the contract a useful comparison with Saku's. Both arrangements enlisted people beyond the worker, but their duties, durations, and provisions differed. They belong within a history of contracted service without becoming interchangeable accounts of what every servant experienced.

Saku's contract survives with unusually explicit provisions for several possible departures. If she fled, Kinnosuke and the guarantors were to find her and return her. If she could not be found, the innkeeper could demand another person or repayment. The document anticipated absence without accepting it as the end of his claim.^[5]

The distinction is practical. Searching for a person and recovering money require different actions and may involve different parties. The contract established an order in which the employer's lost service could be pursued through either. Its financial provision did not make the promise of forced return disappear.

Another clause allowed the innkeeper to move Saku to a different establishment if her service was unsatisfactory or she did not suit him, with recovery of the advance. Movement could consequently preserve a claim over her work instead of releasing her from it. The person deciding to change the workplace was not necessarily the person required to work in the new one.

Then the document addressed marriage. If someone wished to marry her during the term, the arrangement required her consent and the innkeeper's approval of the prospective match.^[6] That clause gave her agreement a specific place. It did not supply a general right to leave the inn for any purpose she chose.

Consent was therefore attached to an act rather than spread evenly across the relationship. Her agreement mattered expressly to the proposed marriage; the transfer clause gave the employer a different kind of discretion. Reading either clause alone would produce an incomplete account of her position.

The document also includes her name and a nail-mark, alongside the seals of Kinnosuke and the guarantors. Its acknowledgment of receiving the fifteen ryō, however, is made in Kinnosuke's voice.^[2] Participation in a written undertaking and receipt of the money it records need not belong to the same person.

This matters to the cost of leaving. Saku's service was committed through a payment that answered another person's immediate expenses. The three-year term connected his financial difficulty with her future time. Nothing in that sequence requires an invented account of what she thought; the allocation of money and obligation is already visible on the page.

The contract does not tell us that she actually fled, changed establishments, or married. Those remain anticipated possibilities. What it preserves is the set of claims that other participants sought to maintain if the expected service ended. Its detailed provisions make departure legible as a problem before supplying any evidence that a departure occurred.

Early Meiji lawsuits change the position from which we see the problem. The worker's departure was no longer merely a contingency in a contract. In the cases studied by legal historian Uno Fumie, women had left employment, and employers wanted courts to help retrieve them.

Uno's published abstract reports an analysis of forty-four lower-court decisions concerning female employees. Many disputes involved demands for their return. The courts rejected forced retrieval on grounds of personal liberty. Her study also identifies a developing distinction between invalid contracts made by fathers or husbands and liabilities that could still be pursued in money.^[7]

These are findings about the decisions in that study, not a claim that every early Meiji employer complied with the same rule. Their significance lies in the remedy being refused. The court could recognize that an employer asserted a loss without authorizing the physical recovery of the woman as the means of repairing it.

Compare the question with Saku's document. The earlier contract connected an unsuccessful search for the worker with alternative provision of a person or money. The Meiji decisions separated a monetary claim from permission to bring the worker back by force. A payment dispute could remain after the claim to compel the person's return had been rejected.

Money could still constrain the future. Ending a power of forced retrieval did not create savings, accommodation, or another wage. The distinction nevertheless changed what an employer could ask the court to do. Financial pressure and authority over a person's bodily return were no longer automatically treated as equivalent ways to obtain performance.

The sequence also gives the women a place in the history of the change. They had already acted by leaving and refusing to return. Litigation followed their refusal. The court's decision addressed an interruption that the employer had failed to settle simply by invoking the contract.

We need not invent a shared motive for those departures. Different women could leave for different reasons, while producing a similar legal question about what would happen next. The surviving judgments bring that question into focus more readily than they recover each worker's full experience.

Employment was being reconsidered through disagreements about its ending. A document signed by relatives could no longer, in the cases Uno discusses, do all the work of establishing the woman's obligation to return. That opened another question for the national Civil Code: which parts of an employment relationship required the worker's own agreement?

The Meiji Civil Code, enacted in the 1890s, addressed one part of that question in Article 625. An employer could not assign the rights arising from employment to a third party without the worker's consent. The worker, in turn, could not send someone else to perform the service without the employer's consent. Unauthorized substitution by the worker could give the employer grounds to end the contract.^[8]

The reciprocity concerned the identities of the parties. A promise to work for one person did not automatically become a promise to work for someone else. Likewise, an employer who had engaged a particular worker had not necessarily agreed to receive any substitute who happened to arrive.

During drafting, Hozumi Nobushige explained this personal character of the relationship. Ume Kenjirō's later commentary made the point through the difference between serving one employer and serving another. The identity of the person on each side belonged to the undertaking itself.^[9]

This produced a revealing contrast with the transfer authorized in Saku's contract. That earlier document expressly gave the innkeeper a way to move her elsewhere if dissatisfied. Under the code's rule, an employer could not rely simply on owning an assignable claim to service. The worker's consent was required to transfer that claim.

But a right to resist a change of employer was not the same as an unrestricted right to quit the existing one. Article 625 protected the identity of the relationship. Other provisions had to say when it could end. Keeping those questions separate prevents consent from becoming an all-purpose answer that the legal text itself did not provide.

The worker's position also remained different from that of a supplier sending an interchangeable product. A personal undertaking could restrict the employer's ability to transfer it while restricting the worker's ability to delegate it. The same principle protected an identified relationship and kept the duty of performance attached to an identified person.

The next question was temporal. When the parties had not specified an endpoint, how could one of them bring the continuing relationship to a close?

The drafting discussion of what became Article 627 exposed a difficulty hidden inside a general rule. For employment without a fixed term, the draft allowed either party to give notice of termination. It also related the notice requirements to the period by which remuneration was determined.^[10]

That sounded orderly until the period became a year. At the commission's ninety-eighth meeting, Hasegawa Takashi challenged the suggestion that annual remuneration was uncommon. In the countryside, he argued, it was familiar for farm servants. Requiring an excessively early notice could conflict with practice. Hozumi acknowledged the objection, and the discussion was postponed.

At the following meeting, the draft included an additional provision: where remuneration was fixed by a period of six months or more, three months' notice would be required. Hozumi explained the proposed interval in terms of time for an employer to find a replacement and for a worker to find another place. These were the drafters' expectations, not measured outcomes for every farmhand.^[11]

The final article retained a two-week default for employment without a fixed term, alongside provisions tied to remuneration periods. Where pay was fixed by a period, notice ordinarily concerned the following period and had to be given during the first half of the current one. For periods of six months or more, the three-month provision applied.^[12]

It would be misleading to compress this into “workers could leave after two weeks.” The applicable arrangement depended on how the employment and remuneration had been specified. The text distinguished the duration of the engagement from the period used to calculate its pay. An annual rate did not necessarily establish an employment contract that ended after one year.

The drafting debate gives that technical distinction a human setting. A rule intelligible to someone thinking chiefly about one kind of employee could operate differently for a rural servant. The commissioners were deciding how long either side should have to prepare for a change it had not initiated.

Notice made the interval itself part of the settlement. One party could begin termination without obtaining a new agreement to end the relationship, but the effect would not always be immediate. The other's expectation of continued work or continued income received a defined period rather than an unlimited claim.

That is a different way of organizing departure from a contract that enlisted guarantors to find and restore a missing worker. The comparison does not erase every later constraint on employment. It identifies a change in the question the governing rules answered: from securing a promised person's continued service toward specifying how an ongoing contractual relationship could be brought to an end.

There was no single historical answer to when a worker could leave. The early Tokugawa limits made duration consequential while leaving existing dependencies in place. The later service contracts supplied endpoints and guarantees, but also gave employers claims if performance stopped. Early Meiji litigation tested whether those claims could reach the departing person's body. The code distinguished consent to transfer from notice to terminate.

The most revealing evidence often begins after the worker is no longer where the employer expects. A search clause, a demand for repayment, a rejected request for forced return, or a notice period identifies what the other party believed should survive the departure.

Saku's contract joined her future labor to money received by someone else. The later courts and codifiers examined here separated claims to service from claims over the worker, and specified remedies for a broken or ending relationship. Those distinctions could matter without making money, family obligations, or the need for another livelihood disappear.

To understand the freedom to leave, therefore, ask what could still be demanded once someone had gone. The answer lies as much in the claim that could no longer be enforced as in the date written for the end of the job.