Skip to content
#small language model Open access

The Gyroscope Is Two Orders Too Small to Explain Why a Bicycle Does Not Fall ── The restoring torque required is 68.3765 N. m; the front wheel's gyroscopic torque is 1.2750 N. m ── A gap of 53.6286 ── The two become equal at 965 km/h ── [Paper 359]

Sep 2026 · Zenodo (CERN European Organization for Nuclear Research)
Vehicle Dynamics and Control Systems

Abstract

A moving bicycle is said to stay up because of the gyroscopic effect of its wheels. This paper asks whether that explanation is the right order of magnitude──the answer is that it is not. No new mathematical theorem and no new law is claimed. Scope of this paper (scope note): No new mathematical theorem and no new law is claimed──the gyroscopic effect, caster trail, and the inverted pendulum are all standard. We do not build bicycle dynamics──the four-degree-of-freedom linearised model and its eigenvalue analysis (weave, capsize) are not entered. They are named and no more. We prove no self-stability──the speed range over which a bicycle is self-stable is not computed. What is computed here is the size of torques. We perform no experiment──no measurement on a real machine is made. The numbers are on parameters stated explicitly. Control is not treated──the rider’s reflexes and learning are not the subject. Tyres are not treated──contact patch deformation, slip angle, and toe are not the subject. Relation to earlier papers: Paper 357 showed maximum power and maximum efficiency to be different points──there there were two objectives; here there are two explanations. Paper 356 showed quantities of the same dimension to be different quantities──here two torques of the same dimension are compared in size. Paper 300 showed whether two things share a root is decidable──by that test, gyroscopic restoration and steering restoration have distinct roots. Paper 353 showed a claim turning false when its condition is dropped──confusing “it acts” with “it suffices” is the same error. What is added is giving both torques as numbers under one set of conditions, showing the ratio to be 53.6286, solving backwards for the 268.14 m/s at which they are equal, giving the time available from the inverted-pendulum time constant, and putting the separator on individual necessity. First, the gravitational torque on an 80 kg bicycle-and-rider leaning 5 degrees is 68.3765 N. m (Section 2). Second, this is the core of the paper. The front wheel’s gyroscopic torque is only 1.2750 N. m, a gap of 53.6286 (Section 2). Third, quadrupling the speed brings it only to 5.1000 N. m (Section 2). Fourth, the two become equal at 268.14 m/s, that is 965 km/h (Section 2). Fifth, what remains is steering: a 5 degree lean is balanced by a turn of radius 29.1385 m (Section 3). Sixth, the separator is whether a mechanism is individually necessary, not whether it acts at all (Section 5). A moving bicycle is said to stay up because of the gyroscopic effect of its wheels──and the gyroscopic effect is certainly real. But it is two orders too small──treating the front wheel as a thin ring (which overestimates its moment of inertia), riding at 5 m/s and steering at 0.5 rad/s gives a gyroscopic torque of 1.2750 N. m. The gravitational torque on an 80 kg bicycle-and-rider with its centre of mass at 1.0 m, leaning 5 degrees, is 68.3765 N. m. The ratio is 53.6286 ── not “slightly short” but nearly two orders of magnitude. Speed does not close the gap──at 20 m/s (72 km/h) the gyroscopic torque reaches only 5.1000 N. m, and the two become equal at v=268.14 m/s, that is 965 km/h. It is no longer a bicycle. What remains is steering──as an inverted pendulum the time constant is 0.3193 s, and falling from 1 degree to 10 takes 0.9558 s, long by the standard of human reaction time. So the rider gets there in time. Recovery means bringing the contact patch back under the centre of mass ── steering into the fall ── and a 5 degree lean is exactly balanced by a turn of radius 29.1385 m. That restoration comes straight out of tantheta=v^2/(Rg), so it is the same order as the gravitational torque. But the strengths of the claims must not be mixed──“two orders too small” is carried by the computation here; “individually unnecessary” is carried by the literature. A design cancelling both gyroscope and trail has been made self-stable, but this paper does not reproduce that experiment. What this paper says independently is the ratio 53.6286, reaching the same conclusion by a different road. One thing separates them──not whether a mechanism acts, but whether it suffices alone, and further whether it is necessary alone. The three are different questions, and a “yes” to the first does not settle the second. Bringing the ratio to 1 would need 53.6286 times the moment of inertia ── a front wheel of some 80 kg ── and at that point the vehicle is not a bicycle. To be said honestly──this paper does not fully answer “then what holds it up”. What is shown is that steering remains as the only candidate of the right order. Self-stability itself is a four-degree-of-freedom eigenvalue problem and does not reduce to a single mechanism. One last thing──“is the effect real” and “does the effect explain it” are different questions. The gyroscopic effect is real, and it is not enough. Stop at confirming that something is real and you are satisfied without ever having looked at the order of magnitude. On the making of this work: The ideas and content of this work stem from the author's own considerations. Assistance from an AI (a large language model) was used for structuring, English translation, and checking the algebra. Any remaining errors or misinterpretations are solely the author's. Feedback and corrections are sincerely appreciated. ----- 走っている自転車が倒れないのは、車輪のジャイロ効果のためだと説明される。本稿が問うのはその説明の桁が合うかである──答えは合わない。新しい定理も法則も主張しない。 本稿の射程(射程注記):新しい定理も法則も主張しない──ジャイロ効果・キャスター角・倒立振り子はすべて既知である。自転車の力学を作らない──四自由度の線形化模型とその固有値解析(ウィーブ・キャプサイズ)には立ち入らない。名前を挙げるにとどめる。自己安定性を証明しない──どの速度域で自己安定になるかは計算しない。本稿が計算するのはトルクの大きさだけである。実験をしない──実車での測定は行わない。数値は明示した諸元の上でのものである。制御を扱わない──乗り手の反射や学習は主題ではない。タイヤを扱わない──接地面の変形・横滑り角・トー角は主題ではない。既刊との関係:論文357 は最大電力と最大効率が別の点にあることを示した──そこでは目的が二つあった。ここでは説明が二つある。論文356 は次元が同じでも別の量であることを示した──ここでは同じ次元のトルク二つの大きさを比べる。論文300 は同根か別根かが判定できることを示した──ジャイロによる復元と操舵による復元は、その基準で別根である。論文353 は条件を落とすと主張が偽になることを示した──「効く」と「単独で足りる」を混ぜると同じ誤りになる。加えたのは、二つのトルクを同じ条件で数に出したこと、比が 53.6286 倍であることを示したこと、並ぶ速度を逆算して 268.14 m/s を出したこと、倒立振り子の時定数から使える時間を出したこと、分離子を「単独で必要か」に置いたことである。 第一に、車体と乗り手 80 kg が 5 度傾いたときの重力トルクは 68.3765 N. m である(第2節)。 第二に、これが本稿の芯である。前輪のジャイロトルクは 1.2750 N. m しかなく、53.6286 倍の開きがある(第2節)。 第三に、速度を四倍にしても 5.1000 N. m にしかならない(第2節)。 第四に、二つが並ぶ速度は 268.14 m/s、すなわち時速 965 km である(第2節)。 第五に、残るのは操舵である。傾き 5 度は半径 29.1385 m の旋回でつり合う(第3節)。 第六に、分離子は「その機構が単独で必要か」であって、「効くか」ではない(第5節)。 走っている自転車が倒れないのは車輪のジャイロ効果のためだ、とよく説明される──そしてジャイロ効果は確かに実在する。だが桁が足りない──前輪を薄いリングとみなし(実際より慣性モーメントを大きく見積もって)5 m/s で走り、操舵角速度 0.5 rad/s でハンドルを切ったときのジャイロトルクは 1.2750 N. m である。一方、車体と乗り手 80 kg・重心 1.0 m が 5 度傾いたときの重力トルクは 68.3765 N. m。比は 53.6286 倍で、これは「少し足りない」ではなく桁が二つ近く違う。速度を上げても追いつかない──20 m/s(時速 72 km)にしてもジャイロトルクは 5.1000 N. m にしかならず、二つが並ぶのは v=268.14 m/s、時速 965 km のときである。もう自転車の話ではない。残るのは操舵である──倒立振り子として見れば時定数は 0.3193 秒、傾き 1 度から 10 度まで倒れるのに 0.9558 秒あり、人の反応時間からすれば十分に長い。立て直す方法は接地点を重心の下へ戻すこと、すなわちハンドルを倒れる側へ切ることで、傾き 5 度は半径 29.1385 m の旋回でちょうどつり合う。この復元はつり合いの式 tantheta=v^2/(Rg) から直接出るので、重力トルクと同じ桁になる。ただし主張の強さを混ぜてはいけない──「桁が足りない」は本稿の計算が支えるが、「単独では必要でない」は文献が支える。ジャイロもトレールも打ち消した自転車が自己安定になる設計が作られているが、本稿はその実験を再現しておらず、文献として挙げるにとどめる。本稿が独立に言えるのは 53.6286 倍という比だけであり、それが別の道から同じ結論に届いている。分けているものは一つ──その機構が「効く」かではなく、「単独で足りる」か、さらに「単独で必要」か。三つは別の問いであって、最初の行で「はい」と言えることが次の行の答を決めない。比 53.6286 を 1 にするには前輪の慣性モーメントを 53.6286 倍にしなければならず、そのとき乗り物は自転車ではない。正直に言えば──本稿は「では何が支えているのか」に完全には答えていない。示したのは、操舵が桁の合う唯一の候補として残ることまでである。自己安定性そのものは四自由度の固有値問題であって、単一の機構に帰着しない。最後に一つ──「その効果は実在するか」と「その効果で説明がつくか」は別の問いである。ジャイロ効果は実在し、しかも足りない。実在を確かめただけで説明を終えると、桁を見ないまま納得してしまう。 作成にあたって:本稿の着想と内容は、著者自身の考察に基づくものです。文章の構成整理や英訳、数式の確認には AI(大規模言語モデル)の助力を得ました。最終的な内容の解釈や誤りがあれば、それらはすべて著者の責に帰します。お気づきの点があれば、ご教示いただければ幸いです。

View source

Similar papers

#small language model Dataset Open access Oct 2026

Socratic guiding questions in synthetic arithmetic data: matched LoRA runs (revision v2)

Supporting data, adapters, predictions and code for the article *Low-Cost LoRA Fine-Tuning of Small Language Models for Multi-Step Arithmetic Reasoning* by Jake O'Grady, Asena Isik Gürhan, Chee Fong Ting and Effirul Ramlan (University of Galway). We generated 20,000 GSM8K-derived arithmetic problems with step-by-step s...

O'Grady, Jake, Gürhan, Asena Isik, Chee, Fong Ting et al. · 465 citations
#computer vision Open access Jun 2016

Software Development in Startup Companies: The Greenfield Startup Model

The results are packaged in the Greenfield Startup Model (GSM), which explains the priority of startups to release the product as quickly as possible, and the need to shorten time-to-market, by speeding up the development through low-precision engineering activities.

Carmine Giardino, Nicolò Paternoster, M. Unterkalmsteiner et al. · 178 citations · ⚡14
#computer vision Open access Oct 2016

Software Startups - A Research Agenda

Software startup companies develop innovative, software-intensive products within limited timeframes and with few resources, searching for sustainable and scalable business models.

M. Unterkalmsteiner, P. Abrahamsson, Xiaofeng Wang et al. · 157 citations · ⚡17
#machine learning Review Open access Oct 2016

“Failures” to be celebrated: an analysis of major pivots of software startups

This study conducts a case survey study based on the secondary data of the major pivots happened in 49 software startups, and demonstrates that customer need pivot is the most common among all pivot types.

Sohaib Shahid Bajwa, Xiaofeng Wang, Anh Nguyen-Duc et al. · 127 citations · ⚡15
#computer vision Review Open access May 2015

A survey study on major technical barriers affecting the decision to adopt cloud services

The comparison of adopter and non-adopter sample reveals three potential adoption inhibitor, security, data privacy, and portability, which underlines the importance of the technical and security perspectives for research investigating the adoption of technology.

Nattakarn Phaphoom, Xiaofeng Wang, S. Samuel et al. · 111 citations · ⚡8
#computer vision Open access Feb 2018

Lean Internal Startups for Software Product Innovation in Large Companies: Enablers and Inhibitors

This study investigates how Lean internal startup facilitates software product innovation in large companies and identifies its enablers and inhibitors, and shows the potential of the method-in-action framework to investigate the Lean startup approach in non-startup context.

Henry Edison, Nina M. Smørsgård, Xiaofeng Wang et al. · 78 citations · ⚡6

Related blog posts

MIT News · Artificial Intelligence Oct 15, 2026

AI model decodes the language cells use to communicate

A new computational approach reveals how cells communicate to assemble into tissues. This could accelerate stem cell engineering, improve disease models, and inform therapies.

Microsoft Research Blog Oct 15, 2026

Proof-based general-purpose computing with Magnetar

Zero-knowledge proofs can verify computing results without rerunning workloads or exposing sensitive data. Magnetar helps bring them into the mainstream with support for standard programming languages, major speed gains, and quantum-resistant security. The post Proof-based general-purpose computing with Magnetar appeared first on Microsoft Research.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.