Sixty days, eight checkpoints: the tempo of a fundraise60 天八节点：融资过程的节拍

1. Proposition: Relationships Are the Ticket In, Cadence Is the Degree of Completion

Relationships decide who receives you; cadence decides whether this round closes.

First, give the opposing view its due. "It comes down to relationships" does hold at the entrance: whom you know decides how soon the first meeting can be scheduled, and what assumptions the other side brings to the table.

But extend the observation a little further along, and the explanatory power drops. The same set of relationships: some rounds close in sixty days, others drag on for ten months; sometimes everyone on the list has been met, and still nobody is willing to move first. Relationships here are a constant, and a constant cannot explain a variable.

The variable is order and time. A financing round consists of only a handful of tasks — put the facts in order, build the materials, calibrate expectations, organise the lead investor, support due diligence, coordinate the investment committee, negotiate terms, complete the closing. None of them is mysterious on its own. The difficulty is that they must happen in a certain order, and land inside the same window. Get the order wrong, and the same amount of work does not buy the same result; let the timing scatter, and the parts already finished expire.

This is what the paper means by cadence: not speed, but sequence and interlock.

---

2. The Eight Checkpoints: Source Text and Formulation

Eight checkpoints across sixty days is seven and a half days each; but the first eighteen days decide how hard the remaining forty-two will be.

The section "60-Day Benchmark Execution Cadence" of Glacier Institute states the checkpoints publicly as follows [1]:

D01 fact base → D09 professional materials → D18 market expectation calibration → D26 organising the lead → D35 deep diligence → D43 investment-committee coordination → D52 core terms → D60 closing （「D01 事实底座搭建 → D09 专业材料开发 → D18 市场预期校准 → D26 领投机构组织 → D35 深度尽调支持 → D43 投委会过会协同 → D52 核心条款谈判 → D60 资金交割收尾」）

The same section also states: "The eight nodes are the only official external wording; the actual schedule is adjusted dynamically according to the company's readiness, the market window and the complexity of the deal" (「八个节点为官方唯一的对外表述；实际排期依据企业准备度、市场窗口与交易复杂度动态调配。」). The same formulation also appears in the public fact file [2] and on the FAQ page [8]. Figure 1 shows where this section sits within the method system published on the website.

First, sixty days is a typical framework, not a promise. The source text's own words are: "Sixty days is not a process chart; it is one push with a close — the nearer the closing, the heavier the moves" (「六十天不是流程表，是一次有收口的推进——越接近交割，动作越重」) [1]. Sixty days is not a squeeze on speed; it is eight things locked into one window [3].

Second, eight checkpoints spread across sixty days is seven and a half days each. That sounds even. It is not. From D01 to D18 takes eighteen days, under three-tenths of the run; yet most of the questions challenged again and again later come out of those eighteen days [3]. This unevenness is the pivot of the whole paper: it decides which checkpoints can be compressed and which collapse the moment they are.

The source text also has a coarser, calendar-level formulation — restore the facts, calibrate the market, shape the table, cross the decision, complete the close [1]. This paper uses only the eight-checkpoint numbering and does not mix the two; Figure 2 shows how the two granularities are drawn.

---

3. Checkpoint by Checkpoint: the Judgement, the Failure Mode, Why It Cannot Be Skipped

The fact base is not a data pack; it is "one set of facts" — the only place in the round where correcting an error is free.

3.1 D01 fact-base construction

Judgement: whether every sentence afterwards stands on the same set of facts [3].

The product of this step is not a data pack. It is consistency. What is judged is not "are the materials sufficient" but "is there a second version" — the same number, in the materials, in the contracts and in the founder's own words: is it the same number?

Failure mode: three versions of the facts are in the room at once, and nobody is lying. The founder holds the technology piece, the institutions hold the market piece, the existing shareholders hold the history piece; every piece is true, and the whole picture cannot be assembled [1]. The second is to read D01 as "collecting documents", so it is handed to the most junior person, or simply folded into D09.

Why it cannot be skipped: the cost gradient. A wrong statement corrected before launch costs roughly nothing; left until D35, changing one page becomes changing a whole timetable [1]. Skipping D01 does not save eight days. It moves those eight days of work to the most expensive place on the timeline.

3.2 D09 professional materials development

Judgement: whether it can converge on one sentence [4].

The standard for materials, in the source text's words, is coherence, not completeness — every number and every chart has to survive being checked backwards [4]. The work at this layer is, plainly, producing evidence: complete the chain and let the other side reach the conclusion themselves.

Convergence is the real difficulty. Construction means subtracting until one sentence is left. That sentence is not an abbreviation; it is the reason a particular buyer acts now. If it cannot converge, you are still talking to everyone — which is talking to no one [4].

Failure mode: thickness in place of coherence. A two-hundred-page deck is usually not a sign of much content but of nobody daring to delete — every page answers some question from some meeting, and the materials become a sediment of questions. The second is drawing the conclusion for the other side: the fuller the conclusion is written, the further the other side steps back.

Why it cannot be skipped: every later meeting must use the same source text. The materials are not for the first meeting; they are for the seventh — by then you are no longer in the room, only the materials are.

Takeaway | The standard for materials is coherence, not completeness; only once you have subtracted down to one sentence do you know who the buyer is.

3.3 D18 market-expectation calibration

Judgement: where the company's price expectation falls within the real valuation band.

The valuation band framework of Glacier Institute splits this into three segment propositions: US$0–1B answers why this team, this technical path, this product form and this moment in the market; US$1–10B calls for benchmark customer validation, real growth data, a diversified shareholder base and milestone-linked rounds; US$10–100B is about global governance architecture, platform ecosystem logic, and the capacity to meet the global public markets [1][6]. The purpose of the framework is to make price, round size, dilution, milestones, investor type and the company's carrying capacity for the next stage match one another (an analytical framework, not a valuation commitment) [1].

Calibration is a judgement about this matching relation: the same company walking into different valuation bands must answer entirely different questions. Set the price one band higher and the questions change to a different set, while the materials still sit in the band below. The external constraint is also clear — existing empirical work shows that fund-level capital inflows significantly affect valuation levels in private markets [15]: the reference frame for price expectations itself moves with the funding environment.

Failure mode: turning calibration into "quote a price first and test the market's reaction" — which amounts to running, on the most expensive firms on the list, an experiment that should have been completed off the list. The second is misreading signals: taking politeness for interest, silence for refusal. The third is calibrating only after a rejection, when a price cut already means something else.

Why it cannot be skipped: calibration is cheap only if it is done before the formal list goes out. Once it has gone out, the expectation becomes an established fact. Lowering it needs a reason, and any reason will be read as "something has happened".

Takeaway | Calibrate before you ask, not after you are refused — every change to an expectation already sent out costs one round of explanation.

3.4 D26 lead-investor organisation

Judgement: who is both willing and able to write the price down.

The source text gives this day the heaviest weight: everything before D26 is laid down for that one day; the lead investor is the first party willing to write the price down [3].

Why is pricing so hard? Nobody wants to price alone: whoever moves first carries the risk of getting the price wrong on their own, so everyone waits for someone else to move. But once one sufficiently credible firm names a price, the question in front of everyone else changes: from "is it worth it" into "is there any allocation left", and the cost of deciding drops [3][4]. This is consistent with the classic models of information cascades: later movers, having observed the actions of earlier ones, decide on those actions rather than on their own private signals, and a queue forms quickly [16][17]. Early research also found that venture capitalists tend to invite peers of similar experience to co-invest in first rounds [13].

So this step is not about persuading more people. It is about finding the one who is willing and able to price. The first duty in investing is safety, not odds [4] — which is why casting a wide net is counter-productive at this checkpoint.

Failure mode: persuading everyone at once. The list is wide, every firm is "following up", and none is pricing. The second is mistaking reputation for effectiveness: a mid-sized firm with a single standard and a short decision chain can run the whole course in two weeks; reputation is one thing, whether it works inside this window is another [3]. The third is letting information arrive out of step — the same set of facts reaching different people in different weeks, so the table never takes shape [4].

Why it cannot be skipped: this checkpoint is the phase transition of the whole timeline. Before it you are producing consensus; after it you are allocating. Without a pricer, the next three checkpoints hang in the air — due diligence has no baseline, the committee has no price, the terms have no anchor.

Takeaway | What you are looking for is not many people but the one who is willing and able to price; once there is a price, the question changes from "is it worth it" to "is there any allocation left".

3.5 D35 deep due-diligence support

Judgement: which questions will blow up at the farthest, least recoverable point.

Due diligence is not answering questions. It is anticipating the distribution of questions. The "Seals" item (「密封」) in Glacier Institute's "Five Checks Before Launch" says exactly this: every small leak in equity, terms, governance or information rights will be discovered only at the farthest point of the voyage (「股权、条款、治理与信息权上的每一处细小漏气，都会在最远的地方才被发现」) [1]. This checkpoint opens the leaks deliberately, at a controllable distance.

What the counterparty is doing has already been profiled: empirical study of investors' internal analysis documents shows that institutions systematically distinguish internal, external and execution risk, and map those judgements onto subsequent terms [10]; a large-sample survey of investors also shows that due diligence and post-investment management are among the most labour-intensive parts of their process [11]. Due-diligence questions are therefore not random. They can be anticipated.

Failure mode: producing materials on the spot. The request list arrives, data are made to order, and what comes out does not match D09 — a second version of the facts appears, and the cost of a D01 left unclean arrives on this day with interest. The second is different answers in different rooms: the same question answered by the technical side, the finance side and the founder in person, with three different precisions. The third is treating due diligence as defence, so every question is taken as an attack.

Why it cannot be skipped: due diligence never tests the materials. It tests D01. Skipping it does not make the problems disappear; it only moves their discovery into someone else's meeting room.

Takeaway | Due diligence tests D01, not the materials; opening a leak yourself costs an order of magnitude less than having it found.

3.6 D43 investment-committee coordination

Judgement: what the counterparty's real internal decision path is, and who will speak for you.

What is special about this checkpoint is that the decision happens in a room where you are absent. Glacier Capital describes this as "translating value into language" (「价值语言转译」): turning a company's real hard value into the standard language that investment institutions and their investment committees can decide on efficiently [1]. What is translated for is not the individual investor but the decision procedure; surveys of investor decision processes likewise show that evaluation and decision inside an institution are highly institutionalised, and cannot be overridden by the personal preference of the person handling the deal [11].

So there are two layers to judge: the content layer — which evidence is recognised in that room; and the process layer — how short this institution's decision chain is, how many gates remain before the meeting, and whether another project is competing for the same agenda this week.

Failure mode: serving only the person you have met. What he carries back is his version of your case, and the challenges he will face are ones you have never heard. The second is treating the committee as a date rather than a process — only on the day do you learn that materials are missing, and once they are added the next meeting is three weeks away, so a slice of the window is eaten. The third is using the same materials for every institution, regardless of each one's decision habits.

Why it cannot be skipped: this is the only checkpoint in the whole process that you cannot execute in person. All preparation must be finished in advance.

Takeaway | You are not in the room on committee day, so before then you must put the words into the hands of someone who can speak for you.

3.7 D52 core-term negotiation

Judgement: which terms move the price, and which terms move the room to choose in the next round.

This checkpoint is the easiest to reduce to "negotiating the valuation". But Glacier Capital's formulation of deal structure is one designed around price, allocation, roles, sequence, terms, and industrial synergy — price is only one of six [1]. This is consistent with the direction of empirical contract research: the core of a venture capital contract is the separate allocation of cash-flow rights, control rights, liquidation rights and board seats, and these rights can be separated from one another [9]; staged financing tied to milestones is itself a governance tool, not merely a payment arrangement [12].

The criterion is the line from the 4D model: every move today changes the option space of next week, the next round and the next stage [5]. A term that looks harmless in this round may become a veto in the next.

What makes a late-stage deal hard is not the valuation but the number of interests inside it. Existing shareholders wanting out, the option pool, leftover performance ratchets, disagreements between management and finance — each additional party is another veto; the workload does not grow linearly with the amount, it grows combinatorially with the number of parties [4].

Failure mode: rules set afterwards. The strategic investor wants one condition, the financial investor wants another, the existing shareholders want a third; each is reasonable on its own. But with everyone naming their own terms, sixty days of work turns into ten months. The method is to set the rules first, write them on paper, the same for everyone — a fair deal environment is in fact an accelerator. Will someone be unhappy? Yes. But he knows on what basis [1].

The second is using the closing date to force concessions. Once a deal starts, every incentive at the table points at closing; admitting the structure is wrong at that moment means fighting everyone's calendar, and nobody finds it easy to [1]. This is precisely why a role that carries no pressure to close needs to be in the room.

Why it cannot be skipped: terms are the only thing that stays in force after this round ends. Price affects only this round; terms affect every round after it.

Takeaway | What is negotiated is not this round's price but the next round's room to choose; setting the rules first is faster than winning each clause.

3.8 D60 closing and funds transfer

Judgement: whether every signature and every follow-on obligation has landed on a specific person.

Glacier Capital breaks full mandate into six dimensions: Trust, Information, Narrative, Tempo, Relationships, Accountability — and the last, Accountability, is defined as: from coordinating terms to managing the close, responsibility rests with named individuals (「条款协调到交割管理，责任落到人」) [1]. The judgement at the closing checkpoint is not "has it been signed" but "who is keeping watch".

Failure mode: treating signature as completion. Documents scattered across people's chat windows, signing dates left blank, versions that do not match, renewal texts that exist but were never countersigned. None of these is a problem on the day; they surface in the next round — when the next round starts, D01 has to be done all over again, because the previous round's source documents can no longer be assembled. The second is disbanding the project team the moment the closing is done, so nobody takes over the follow-on obligations.

Why it cannot be skipped: a line about the Amundsen discipline in the source text can be carried over directly: arrival is not the end; returning to camp is. Nor is closing the end: only when the company still stands in the next cycle is the journey complete (「抵达不是终点，回到营地才是；交割也不是终点，公司在下一个周期还站得住，这一趟才算走完」) [1]. The cost of skipping the close is not paid in this round. It is paid in the next.

Takeaway | Signed is not closed — this checkpoint is done only when responsibility has landed on a person.

---

4. Why Cadence Matters More Than Effort

Effort raises output at a single point; cadence decides whether those outputs arrive in the same window at the same time.

First, the cost gradient. The same error carries a different price at different checkpoints, and the gap is an order of magnitude. Effort can raise the quality of output at a single point, but it cannot change when an error is discovered — only cadence can. That is the whole reason for the discipline that "most of the work belongs before launch" (「绝大部分工作放在发射之前」) [1].

Second, the concurrency structure. A financing round is never single-threaded. Glacier Capital's 4D formulation: one dimension is linear communication with a single firm and a single decision pushed forward; two dimensions is managing several investors, multiple strands of material and due diligence processes at once; three dimensions is coordinating the company itself, existing shareholders, new capital, industrial partners, valuation expectations and market opinion; four dimensions is introducing the time dimension and working out how today's decision cascades into later rounds, the capital path and long-term strategy [1][5]. In a concurrent system, local acceleration is not overall acceleration — a thread that runs too fast only collides sooner with another that is not yet ready.

Third, the window is exogenous. The real constraint is not in the company, it is in the window, and windows are counted in weeks, not quarters [1][3]. This is not merely industry self-description: early empirical work found that professional investors push portfolio companies into the public market when market valuations are high [14]; the evidence that capital inflows affect valuations shows that the window at the entry end is exogenous as well [15]. Effort decides how much you do inside the window; it does not decide when the window opens — so only one thing can be done: make sure that when the window opens, all eight things can arrive at once.

There is also a companion practice: eighty per cent of the time goes to the material that is most urgent now, twenty per cent is held for the next bottleneck [1]. Putting everything into the present looks maximally efficient, but when the next gate opens the material is still a blank page; that twenty per cent is what makes the next stage's material half-built by the time it is needed. The four levels of the 4D formulation are shown in Figure 3.

---

5. When the Cadence Does Not Hold

When the cadence does not hold, the first thing to move is not the schedule; it is to admit which checkpoint has been determined exogenously.

A framework that reports only the conditions under which it holds, and never the conditions under which it fails, is an advertisement, not a method. In the following five situations, the eight-checkpoint cadence either partly fails or does not apply at all.

One, the very early stage. When the company does not yet have a valuation band that can be calibrated, D18 loses its reference — what this segment must answer is why this team, this technical path, this product form and this moment in the market [1][6], and how fast those propositions can be answered depends on the diffusion of understanding, not on scheduling. Here the order of the eight checkpoints still holds; the time scale does not: D01 and D09 may take up most of the cycle, and D26 waits for an event that cannot be scheduled. Do not impose sixty days' density on a project whose consensus is still forming.

Two, the round already has a dominant investor. When an existing shareholder or a single institution in fact dominates the round, D26 is determined exogenously — the pricer already exists and does not need to be organised. Research on staged financing suggests that incumbent investors already carry the role of continuous monitoring and follow-on decisions [12]; to go on "organising the lead" on the original cadence would be read as distrust of the incumbent. The centre of gravity of the whole timeline shifts later: the work concentrates in D52 and D60, and the earlier checkpoints reduce to information alignment.

Three, the window has not opened. When the window has not come, the cadence is not executed; it is shelved: nothing is worth more than turning the facts over until they are clean. Glacier Capital states this explicitly — when the window is closed, sit patiently, even ten months on the bench with one company; when it opens, close in dense succession (「窗口期未至时耐得住寂寞，可以陪一家公司坐十个月冷板凳；窗口开启时迅速推进多轮密集交割」) [1]. D01 can be stretched indefinitely, and the checkpoints after D26 should not be started — starting them now is opening the show with no audience.

Four, the number of parties crosses a threshold. In late-stage deals, mergers and restructurings, and multi-entity reorganisations, what is hard is not the valuation but the number of interests, and the workload grows combinatorially with the number of parties [4]. Past a certain number of parties, the sixty-day scale loses its meaning. The eight checkpoints remain a valid checklist, but no longer a schedule.

Five, the fact base itself is not clean. If there are unresolved issues in the corporate history, the shareholding structure or key contracts, this round does not start at D01. It starts before D01. Forcing D01 to be the starting line only carries the problems into every later checkpoint. The line in the source text, "it depends on how clean the fact base is" (「要看事实底座有多干净」), answers exactly the question "which day your round should start from" (「这一轮该从哪一天算起」) [3].

These five situations are not "exceptions". They are the boundary of application of this cadence. A framework that cannot answer "within what range does it hold" has dropped a dimension.

---

6. What This Cadence Shows

A checkable process is harder to fake than a result; being open to rebuttal is what makes a framework sincere.

The eight checkpoints are not packaging for a record of results. They show a process capability that a third party can check.

Glacier Capital writes its demands on itself as the four lines of the "Glacier Quality Wall". The first is "testable fundraising efficiency: the process traceable, the milestones reconcilable" (「可被检验的融资效率——过程可回溯，节点可核对」); the other three concern transaction quality, economy of the management team's time, and reducing noise in the market [1][7].

The first line is exactly the reason the eight checkpoints exist. "The process traceable, the milestones reconcilable" (「过程可回溯，节点可核对」) means the cadence is designed to be open to challenge: every checkpoint has a defined product, and the product can be set against the judgement made at the time. Paired with it is the review system — a brief review after every meeting, a full one that same day, a weekly synthesis, a complete monthly retrospective; the review is a metronome, not a fire brigade [1].

Results are easy to narrate; a process is hard to fake. Looking only at results cannot answer "what would happen if it were run again"; only laying the process open leaves an entry point for rebuttal. This paper therefore reports no transaction results.

---

7. Conclusion

Sixty days is not a marketing number. It is a schedule. It is not a squeeze on speed; it is eight things locked into one window [3]. Slow is fine; scattered is not. Fast and slow are both approaches; scattered is the accident [3].

As for which day counts as D01, the answer has been given above: it depends on how clean the fact base is.

---

Eleven papers are not eleven topics. They are eleven links on one chain: segment first, then sequence, then take apart the one move in each link that most often goes wrong.

Read nextGI-WP-2026-P12　Losses at the Handoff: Where the Boundary of a Full Mandate Should Be Drawn八个节点每换一次手就损耗一次。P12 是 P2 的成本面。

FOR FOUNDERS

For founders这一轮怎么打、要多久、我们接不接，一次说清。进去看 →

转人工 · ASK A HUMAN把这个问题交给赵羚茜（Gogo Zhao，投资人关系负责人）安排——她会把它交到对的人手里。 联系庚辛

庚辛人生（Certainty of Money） · 庚辛研究院通讯论文 · GitHub · PDF 全文 · DOI · 可引用，CC BY 4.0。

References

Style: Author (Year). Title. Journal/Publisher, Volume(Issue), Pages. DOI or accessible link. Grouped by source grade: primary materials are Glacier's public statements; secondary literature is peer-reviewed or working papers.

Primary materials (Glacier's public statements; the checkpoint names and definitions are taken from [1] and have not been reworded in this paper)

[1] Glacier Institute (2026). 庚辛研究院官网（含「60 天标杆执行节奏」「0—1000 亿美元估值带分析框架」「庚辛质量墙」「发射前五检」「全托管六维」各节） [Glacier Institute website (including the sections "60-Day Benchmark Execution Cadence", "0–100B USD valuation band framework", "Glacier Quality Wall", "Five Checks Before Launch" and "the six dimensions of full mandate")] [in Chinese]. Glacier Capital official website. https://glacier.mba/institute.html (accessed 2026-09-20)

[2] Glacier Capital (2026). AI 可读事实档案 llms.txt [AI-readable fact file llms.txt] [in Chinese]. Glacier Capital official website. https://glacier.mba/llms.txt (accessed 2026-09-20)

[3] Glacier Capital (2026). 六十天，是一张排期表 [Sixty Days Is a Schedule] [in Chinese]. Certainty of Money, essay No. 19. https://glacier.mba/essays/p16-sixty-days.html (accessed 2026-09-20)

[4] Glacier Capital (2026). 由外而内，拧紧四次 [From Outside In, Tightened Four Times] [in Chinese]. Certainty of Money, essay No. 20. https://glacier.mba/essays/p17-vortex.html (accessed 2026-09-20)

[5] Glacier Capital (2026). 今天的动作，决定下周还剩几张牌 [Today's Move Decides How Many Cards Are Left Next Week] [in Chinese]. Certainty of Money. https://glacier.mba/essays/p14-4d-time.html (accessed 2026-09-20)

[6] Glacier Capital (2026). 先分类，再谈价 [Classify First, Then Talk Price] [in Chinese]. Certainty of Money. https://glacier.mba/essays/p15-price-bands.html (accessed 2026-09-20)

[7] Glacier Capital (2026). 自有资金与产业资本 [Own Capital and Industrial Capital] [in Chinese]. Glacier Capital official website. https://glacier.mba/capital.html (accessed 2026-09-20)

[8] Glacier Capital (2026). 常见问题 [Frequently Asked Questions] [in Chinese]. Glacier Capital official website. https://glacier.mba/faq.html (accessed 2026-09-20)

Secondary literature (venture capital contracting, process and information cascades)

[9] Kaplan, S. N., & Strömberg, P.（2003）. Financial Contracting Theory Meets the Real World: An Empirical Analysis of Venture Capital Contracts. Review of Economic Studies, 70(2), 281–315. https://doi.org/10.1111/1467-937X.00245

[10] Kaplan, S. N., & Strömberg, P.（2002）. Characteristics, Contracts, and Actions: Evidence from Venture Capitalist Analyses. NBER Working Paper No. 8764. https://doi.org/10.3386/w8764

[11] Gompers, P., Gornall, W., Kaplan, S. N., & Strebulaev, I. A.（2016）. How Do Venture Capitalists Make Decisions? NBER Working Paper No. 22587. https://doi.org/10.3386/w22587

[12] Gompers, P. A.（1995）. Optimal Investment, Monitoring, and the Staging of Venture Capital. The Journal of Finance, 50(5), 1461–1489. https://doi.org/10.1111/j.1540-6261.1995.tb05185.x

[13] Lerner, J.（1994）. The Syndication of Venture Capital Investments. Financial Management, 23(3), 16–27. https://doi.org/10.2307/3665618

[14] Lerner, J.（1994）. Venture Capitalists and the Decision to Go Public. Journal of Financial Economics, 35(3), 293–316. https://doi.org/10.1016/0304-405X(94)90035-3

[15] Gompers, P., & Lerner, J.（2000）. Money Chasing Deals? The Impact of Fund Inflows on Private Equity Valuations. Journal of Financial Economics, 55(2), 281–325. https://doi.org/10.1016/S0304-405X(99)00052-5

[16] Banerjee, A. V.（1992）. A Simple Model of Herd Behavior. The Quarterly Journal of Economics, 107(3), 797–817. https://doi.org/10.2307/2118364

[17] Bikhchandani, S., Hirshleifer, D., & Welch, I.（1992）. A Theory of Fads, Fashion, Custom, and Cultural Change as Informational Cascades. Journal of Political Economy, 100(5), 992–1026. https://doi.org/10.1086/261849

Verification note: [1]–[8] were verified as accessible by HTTP fetch on 2026-09-20; for [9]–[17], title, journal, volume, issue, pages and year were checked item by item against Crossref metadata, and all DOIs resolve. The set of statements "the Amundsen discipline · three tons of supplies · thirty kilometres a day" (「阿蒙森纪律·三吨物资·每天三十公里」) is attributed on the website to a 2012 speech by Wang Xing (王兴); this paper cites it via [1] and has not separately verified the original source of that speech.

---

Glacier Institute 庚辛研究院. Sixty days, eight checkpoints: the tempo of a fundraise[R]. Glacier Institute Working Paper GI-WP-2026-P2-EN. 2026. DOI:10.5281/zenodo.22869123.

Glacier Institute. (2026). Sixty days, eight checkpoints: the tempo of a fundraise (Glacier Institute Working Paper No. GI-WP-2026-P2-EN). Zenodo. https://doi.org/10.5281/zenodo.22869123

@techreport{glacier2026p2en,
  author      = {{Glacier Institute}},
  title       = {{Sixty days, eight checkpoints: the tempo of a fundraise}},
  institution = {Glacier Institute},
  type        = {Working Paper},
  number      = {GI-WP-2026-P2-EN},
  year        = {2026},
  doi         = {10.5281/zenodo.22869123},
  url         = {https://doi.org/10.5281/zenodo.22869123}
}