The Same Field, Twice

Two weeks of the same form, read the same way, and the middle of it does not move.

Between 5 and 13 August 2026, EDINET's list carried 217 registration statements for domestic investment trust beneficiary certificates from 30 filers. Between 14 and 19 August it carried 176 from 36. Every one of the 393 prints a ceiling on the cover — the largest amount of beneficiary certificates the registration covers — and a subscription period a page or two further in, in 第一部【証券情報】.

In the first week the commonest ceiling was ¥1 trillion (US$6.1 billion), printed on 83 of the 217. In the second it was ¥1 trillion again, on 61 of the 176. The median in both weeks is ¥1 trillion.

The subscription periods hold too. The first week's median span is 183 days, the second week's 184. Both weeks print the renewal wording — that the period is renewed by filing the registration again before it ends — in almost every 【申込期間】 field: 214 of 217, then 173 of 176.

That is three measurements taken a week apart on different documents from partly different filers, and they land in the same place. A form that asks each filer to choose a number gets the same choice more often than any other — 83 of 217, then 61 of 176.

The edges are a different matter.

The first week printed 18 distinct ceiling values, the second 13. Seven values appear only in the first — ¥2 billion (US$12.2 million), ¥250 billion (US$1.53 billion), ¥700 billion (US$4.27 billion), ¥9 trillion (US$54.9 billion), ¥20 trillion (US$122 billion), ¥30 trillion (US$183 billion) and ¥40 trillion (US$244 billion) — and two only in the second, ¥30 billion (US$183 million) and ¥6 trillion (US$36.6 billion). The mode is stable; the tail is not.

Summing the largest figure printed on each cover gives ¥914 trillion (US$5.58 trillion) for the first week and ¥519 trillion (US$3.17 trillion) for the second. Neither is a quantity of anything. They are sums of self-set maxima, and a sum of self-set maxima is dominated by the largest of them: ten covers in the first week printed ¥20 trillion (US$122 billion) or more, and those ten account for ¥250 trillion (US$1.53 trillion) of the ¥914 trillion.

The subscription periods spread differently as well. In the first week 185 of the 217 fell between 180 and 186 days, and 24 ran 350 days or longer. In the second week 124 of the 176 fell in that same band and 41 ran 350 days or longer.

The largest single change is in a field that carries no number at all.

【縦覧に供する場所】, the place where the document is made available for inspection, is answered with a form of 該当事項はありません on most covers and with the name of an exchange on the rest. In the first week two of the 217 named the Tokyo Stock Exchange. In the second, fourteen of the 176 did.

The fourteen are exchange-traded funds, and they come in runs. Norinchukin Zenkyoren Asset Management filed four, all NZAM 上場投信 — one on US Treasuries of seven to ten years, one on German government bonds of the same range, one on French government bonds of the same range, one on the DAX. Daiwa filed three, Nomura three, Global X Japan two, and Sumitomo Mitsui DS and Nissay one each.

An exchange-traded fund registers the same way an ordinary one does, on the same form, with the same ceiling field and the same period field. The inspection line is the field this piece counted, and it moves because a listed fund has an exchange to name. Whether anything else on the form moves with it is not measured here.

So the two weeks agree about what a ceiling usually is and disagree about how many listed funds happened to register. The first is a property of the form. The second is a property of the fortnight.

On the counts. Both figures are counts of documents rather than of list rows: EDINET lists some documents on more than one date, and each week's population here is the set of distinct document IDs whose description begins 有価証券届出書（内国投資信託受益証券）. The second week's index was pulled on 23 August, after it closed; the first week's was read on its last day, so a later pull of that week could return more than 217.

The ceilings are read from the cover field by the same extraction in both weeks, taking the larger where a cover prints a per-fund and a combined figure and the continuing one where it prints an initial and a continuing figure. Applying the opposite rule changes both totals and does not change either mode.

What this does not establish: that the second week's shape would hold in a third. Two weeks is two observations, and the field that moved most between them moved because of which filers arrived, not because of anything the form did.

On currency: the yen figures above are converted at ¥163.8 = US$1, the single rate this file uses, taken for 24 July 2026, for which the daily reference rate is 163.82.