Tilted Scales: Special Courts, Protective Statutes, and the Price Paid by the Other Litigant
The ceiling fan in the designated court moved as if it too were tired of the afternoon. A plastic bottle of water sat unopened on the expert’s table. The file was medical, the allegation grave, the complainant a woman whose life had clearly been hard. The presiding officer leaned forward, not looking at the notes, and said to the room—half instruction, half creed—
“Hamari mahila ko kuch nahi hona chahiye.”
Our woman should not come to harm.
The sentence was not in the Code. It was not in the special statute under which the court sat. It was the atmosphere of the room made audible. Once you have heard that sentence, you cannot un-hear how a forum built for speed and protection can stop being a court and become a shelter with a gavel.
This essay is not an argument that women, children, Scheduled Castes, Scheduled Tribes, or consumers invented their injuries. They did not. Ordinary courts in India are slow, expensive, and often deaf. Special laws and special courts were born from that deafness. The question is narrower and more uncomfortable: when the State designs a forum whose moral job is to stand with the “weaker” party, what happens to the person who walks in as the other side—the doctor, the husband, the officer who signed a file, the neighbour named in a midnight FIR? Can a process be just in the large and still unjust in the particular? The record of Indian law says yes. Repeatedly.
I. Why the special court exists
If you sit in a regular civil court long enough, you learn the grammar of delay. Adjournment is a dialect. A medical-negligence suit filed when a child is five may be argued when that child is twenty. A woman beaten in 2010 may still be waiting for framing of charges in 2016. A Dalit family whose house was burned may watch the accused walk the same village road because the trial has not begun.
The political answer was not only to reform the ordinary court. It was to build parallel rooms with different clocks and different presumptions.
Consumer commissions under the Consumer Protection Act, 1986 (now 2019) were meant to be cheap, summary, and sympathetic to the buyer who could not fight a corporation. After Indian Medical Association v. V.P. Shantha (1995), medical services for consideration entered that net. The doctor who had been trained to fear only the medical council suddenly faced a three-member bench that could award lakhs without a full-dress civil trial.
Mahila courts and family courts, and the criminal architecture around Section 498-A IPC (now mirrored in the new penal code), were meant to answer the closed door of the sasural. Arnesh Kumar v. State of Bihar (2014) itself recited the original purpose: a provision to combat cruelty in the matrimonial home.
The Scheduled Castes and Scheduled Tribes (Prevention of Atrocities) Act, 1989, with its special courts and the famous bar on anticipatory bail in Section 18, was written against a history that needs no footnote: humiliation as public theatre, police who would not register, witnesses who would not speak.
POCSO courts after 2012 were built because child sexual abuse disappeared inside ordinary trial rooms—delay, confrontation, and a culture that treated the child as a small adult witness.
Each of these is, in the broad sense, a reply to inequality. Article 15(4) of the Constitution even names protective discrimination. Union of India v. State of Maharashtra (2019), reviewing the SC/ST Act controversy, said the Court must not treat “all of them as a liar.” That sentence has moral weight. A republic that pretends every complainant and every accused starts from the same hill is lying about the hill.
The trouble begins when the reply hardens into a habit: the disadvantaged party is not merely protected by procedure; the disadvantaged party is presumed to be the truth of the case.
II. The consumer commission: a court that already knows who is weak
Walk into a district consumer commission with a doctor. The board is often a judicial member and two others. The language is of “deficiency” and “service.” The complainant need not prove negligence with the precision of a civil suit. Expert evidence is frequently treated as optional. The file is thin. The emotion is thick.
Jacob Mathew v. State of Punjab (2005) tried to draw a line the commissions still step over. The Supreme Court held that a doctor is not criminally negligent for every bad outcome; negligence in crime must be gross; and before a doctor is thrown into the criminal process, a competent medical opinion should ordinarily exist. Kusum Sharma and later benches repeated the Bolam idea: an accepted professional practice is not negligence merely because another school exists, or because the patient died.
Yet the daily life of consumer litigation is different from the Supreme Court’s prose. A stillborn child, a failed surgery, a scan that missed a twin—these arrive as stories of grief. Grief is not evidence. But grief sits in the first chair. The doctor sits in the second with a bundle of indoor papers that look, to a lay member, like evasion.
You have seen this in the Bihar pattern and its opposite. When a gynaecologist trusts a “normal” ultrasound against a woman screaming in pain, a State Commission says the scan is only an aid—Neeraj Kumar Jain, SCDRC Bihar. When another team ignores ascites and pleural effusion on the same kind of scan and treats ureteric colic until the patient dies of pancreatitis, the National Commission says the report was ignored—Maya Sharma v. Raj Hospital. Both results can be defended as correlation. What the doctor experiences in the room is not correlation. It is a forum that starts from the premise that the hospital has the files, the money, and the duty, and therefore the benefit of doubt is a luxury the patient should not have to buy.
Is that “always” pro-consumer? No honest reader of the National Commission’s docket can say always. Doctors win when records are complete and an expert board is kind. They lose when records are messy—and Indian indoor notes are often messy because the ward is overcrowded, not because someone planned a fraud. The tilt is this: the same mess that is routine in a government hospital becomes, in a consumer forum, proof of a guilty mind.
A fair court would say: the consumer is structurally weaker, so we lower cost and speed the listing. An unfair court says: the consumer is structurally weaker, so we lower the standard of proof. The first is reform. The second is a different religion of adjudication.
Dialogue, if you want it as it happens:
“Doctor sahib, patient mar gaya. Aapke paas record hai. Record mein gap hai.”
“Sir, the night notes were written by the intern. The consultant saw the patient at 11.30.”
“Toh consultant late aaya. Hamara consumer ko insaaf chahiye.”
Notice the pronoun. Hamara consumer. The bench has already chosen a family.
III. The matrimonial criminal court: protection that learned to arrest first
Section 498-A was drafted as a shield. By 2014 the Supreme Court was willing to say, in open text, that it had become a weapon in some hands. Arnesh Kumar v. State of Bihar (2014) 8 SCC 273 is not a men’s-rights pamphlet. It is a judgment about the police habit of arresting because arrest is lawful, not because arrest is necessary. Justice C.K. Prasad wrote that arrest brings humiliation, curtails freedom, and casts scars forever. The Court directed that in offences punishable up to seven years, police must apply Section 41 CrPC: necessity, reasons, checklist, magistrate’s scrutiny. The directions were not confined to 498-A; they were written because 498-A had become the exhibit.
Three years later, Rajesh Sharma v. State of U.P. (2017) tried Family Welfare Committees as a pre-arrest filter. That mechanism was later rolled back as judicial overreach into investigation. The pendulum is the point. Every few years the apex court admits misuse; every few years the political weather demands that admission be softened so that the original victim is not “disbelieved.”
Mahila courts and designated women’s courts intensify the weather. They are not illegal. Specialisation can mean better recording of testimony, screens, in-camera hearings. Specialisation can also mean a closed culture in which the accused’s mother, sister, and eighty-year-old father are treated as a single male conspiracy because the statute allows relatives to be named and the social script already has a villain.
The expert who heard “hamari mahila ko kuch nahi hona chahiye” was not hearing a finding of fact. He was hearing a prior. If the woman must not be harmed by the process, then cross-examination that tests inconsistency becomes cruelty. Bail that lets a man return to the same house becomes danger. Acquittal statistics—often high in 498-A because allegations are omnibus and evidence is thin—are read not as a warning about over-criminalisation but as proof that courts are still patriarchal.
Both readings contain a piece of truth. Women are still burned, locked, and broken for dowry and for the simpler crime of existing. Men and their families are still ruined by overnight FIRs in which the only particular is a date of marriage. A court that cannot hold both facts in one head will always tilt. It will tilt toward the story that feels like justice on that afternoon.
Humanise the other chair. He is not always a monster. Sometimes he is a clerk whose wife left after two years and whose parents have been asked to furnish bail they do not understand. Sometimes she is not a strategist. Sometimes she is a woman who was hit and has no other door. The unfairness is not that the law opened a door for her. The unfairness is that the same door became a tunnel in which the other person is not allowed to be a person.
IV. The Atrocities Act: when history enters the bail order
No serious person denies why the 1989 Act exists. Caste atrocity is not a courtroom metaphor. It is a well, a road, a word shouted in a police station that refuses to write.
The legal design, however, is extraordinary. Section 18 originally barred anticipatory bail. Later amendments, especially after the political storm of 2018, restored a hard edge: no preliminary enquiry as a condition for FIR in the manner the Supreme Court had tried to mandate; arrest without the extra filters the Court had invented.
In Dr. Subhash Kashinath Mahajan v. State of Maharashtra (2018) 6 SCC 454, a two-judge bench looked at a public servant caught in a prosecution that the Court itself called an abuse of process. It said there is no absolute bar on anticipatory bail if no prima facie case is made out or the complaint is mala fide. It noted “instances of abuse” by vested interests. It asked for approval before arrest of public servants.
The country answered with protest, not with a seminar. In Union of India v. State of Maharashtra (2019), a three-judge bench recalled parts of those directions as encroachment on the legislature and as contrary to protective discrimination under Article 15(4). “Can’t treat all of them as a liar,” the Court said, in substance. In 2020 the Court upheld the 2018 legislative amendment. Anticipatory bail remains possible, later benches have said, where the FIR does not even disclose a prima facie offence—but the culture of the special court is not the culture of those paragraphs. The culture is: this Act is a civil-rights statute with a criminal body. Doubt is a luxury the historically powerful should not enjoy.
Here the tilt is most honest and most dangerous. Honest, because the Act’s drafters intended asymmetry. Dangerous, because a false case under this Act is not a civil inconvenience. It is stigma that outlives acquittal, suspension from service, and a village that has already held its own trial.
The public servant who refused a sanction, the teacher who marked a student, the neighbour in a land fight that suddenly acquired a caste sentence—these people are not “the caste system.” They are litigants. A republic that cannot distinguish the system from the defendant will keep producing Mahajan-type files and then be shocked when the Supreme Court, for one season, speaks of liberty.
V. POCSO: the child, the presumption, and the man who cannot be ordinary
POCSO is the hardest chapter to write without sounding like a person who has never sat with a child who was hurt. The Act’s object is right. Children are not small adults. Delay destroys them. Confrontation in an open court is a second assault.
The Act also does something ordinary criminal law rarely does so bluntly. Sections 29 and 30 raise presumptions as to culpable mental state once foundational facts are proved. Special courts are directed to record evidence quickly. Identity is protected. The Bombay High Court’s “skin-to-skin” reading in Satish was stayed and then crushed by the Supreme Court in Attorney General for India v. Satish (2021) 4 SCC 712. Sexual intent, not epidermis, is the ingredient. That correction was necessary. A glove or a cloth cannot become a loophole.
Yet the same architecture that saves the child can unmake an adult who should have been tried, not pre-convicted. A school touch that is clumsy is not always a sexual touch. A family custody war can discover a POCSO paragraph because POCSO moves faster than family court and because the stigma does the work of proof. Defence counsel who test a child’s testimony too hard are painted as monsters; counsel who do not test it watch a conviction rise from a single consistent sentence.
The Supreme Court has, in other cases, warned against treating every allegation as the end of inquiry. Those warnings live in law reports. In the special court they live less loudly than the poster on the wall that says the child is to be believed.
Believe the child as a starting discipline of listening. Do not confuse listening with a verdict. That distinction is the whole of criminal justice. When a designated court forgets it, the accused is not “the patriarchy.” He is a particular human being who may be guilty or may be the collateral of a good law.
VI. Are they “always” unfair? No. Are they structurally tilted? Yes.
It is sloppy to say consumer courts always favour consumers, or that every mahila court convicts, or that every SC/ST prosecution is false. Acquittals exist. Doctors win. Husbands get quashing orders under Section 482 when allegations are omnibus. High Courts still grant bail in POCSO when the story collapses.
The tilt is not a 100–0 scoreboard. It is institutional design plus institutional mood.
Design: reverse or eased burden, bar or chill on anticipatory bail, summary procedure, special courts whose only diet is one class of complainant.
Mood: the sentence you heard. Hamari mahila. Our consumer. Our child. Our oppressed.
Once the pronoun is “our,” the other side is not a citizen. He is an obstacle.
In the broader frame of Indian inequality, that mood is understandable. The ordinary court failed too many people for too many decades. Speed and solidarity were the promised cure. What arrived, in too many rooms, was solidarity without the patience of proof.
Fairness to litigants is not the enemy of fairness to groups. It is the only way a protective statute remains law rather than campaign. Arnesh Kumar understood this about arrest. Jacob Mathew understood it about doctors and the criminal process. Kashinath Mahajan, before it was partly undone, understood it about caste statutes and liberty. Each of those judgments was attacked as betrayal. Each was, in fact, an attempt to keep the special forum a court.
VII. A closing scene
The expert packed the file. The fan still turned. In the corridor a woman waited with a child asleep on her shoulder. On the other bench a man in a cheap shirt asked his lawyer, in a voice meant to be private, “Sir, bail hoga?” The lawyer shrugged in the manner of people who have shrugged too often.
Both of them deserve a room where the chair does not already belong to one family.
If we must have consumer commissions, mahila courts, SC/ST special courts, and POCSO courts—and we must, because the alternative is the old delay dressed as neutrality—then the training of those chairs should include one unfashionable sentence to sit beside the other:
Hamare litigant ko bhi kuch nahi hona chahiye jo qanoon ke bina ho.
Our litigant, too, should suffer nothing that the law itself has not first proved.
Until that sentence is as easy to say as the first, these courts will remain necessary in the republic’s story of inequality, and unsafe in the private story of the person who walked in without the right pronoun.
Supreme Court Precedents on Bias
Indian bias law sits on two Latin pillars that the Supreme Court has repeated for seven decades: nemo judex in causa sua (no one may be a judge in his own cause) and audi alteram partem. The Court has never required proof that the decision-maker’s mind was in fact crooked. It asks whether a fair-minded person, knowing the facts, would reasonably fear that the mind was not open. That is why “justice must not only be done but must also be seen to be done” — the Sussex Justices line — is quoted in Indian reports almost as often as it is in English ones.
What follows is the doctrine as the Supreme Court has actually built it: tests, types of bias, recusal, and the limits the Court itself has drawn so that the plea of bias does not become a tool to pick a bench.
1. The tests: from suspicion to “real likelihood” to “real danger”
Manak Lal v. Dr. Prem Chand Singhvi, AIR 1957 SC 425, is the starting point. An advocate faced a professional-misconduct tribunal. One member had earlier appeared against him in related proceedings. The Court held that actual bias need not be proved. It is enough if there is a reasonable ground for assuming the possibility of bias, such as would produce in the mind of the litigant or the public a reasonable doubt about fairness. The Court also planted a second seed: waiver. If the party knew the facts and sat silent, he cannot spring the objection after an adverse result.
A.K. Kraipak v. Union of India, (1969) 2 SCC 262, moved the doctrine from courts into administration. A member of the Forest Service selection board was himself a candidate. He stepped out when his own name was discussed but stayed in the room when rivals were ranked. Hegde J. formulated the sentence still used in every service-matter brief:
The real question is not whether he was biased. It is difficult to prove the state of mind of a person. Therefore what we have to see is whether there is reasonable ground for believing that he was likely to have been biased. A mere suspicion of bias is not sufficient. There must be a reasonable likelihood of bias. In deciding the question we have to take into consideration human probabilities and the ordinary course of human conduct.
Kraipak also held that the wall between “judicial” and “administrative” action does not shut natural justice out of selections, inquiries, and boards. A biased recommendation can infect the later approval of UPSC or government.
Ranjit Thakur v. Union of India, (1987) 4 SCC 611, shifted the camera from the judge’s conscience to the litigant’s eye. A court-martial confirmed punishment after the confirming officer had already shown hostility. The Supreme Court said the proper question is not “Am I biased?” asked honestly by the judge, but whether the party before him would reasonably apprehend that he would not get a fair hearing. Reasonable apprehension of bias is enough; proven prejudice is not required.
Later benches sometimes speak of a “real danger” of bias (the English Porter v Magill / Medicaments vocabulary), especially in the NJAC recusal discussion. In substance Indian law still runs on Kraipak + Ranjit Thakur: not fanciful suspicion, not a demand for a confession of partiality — a substantial possibility judged by ordinary human conduct.
2. Types of bias the Court has named
Pecuniary interest — automatic disqualification
Direct financial interest, however small, disqualifies. No inquiry into “likelihood” is needed. The NJAC Constitution Bench, Supreme Court Advocates-on-Record Association v. Union of India, (2016) 5 SCC 1 / 5 SCC 808 (recusal order), restated this: where the judge has a pecuniary interest in the outcome, disqualification is automatic. Justice S.H. Kapadia’s recusal from a Vedanta/Sterlite matter because of a shareholding is the public illustration of the same rule.
Personal bias
Friendship, kinship, prior professional alignment, or being a rival for the same post (Kraipak). J. Mohapatra v. State of Orissa, AIR 1984 SC 1572: an author sitting on a textbook-selection committee that was considering his own book. Possibility of bias could not be ruled out.
Rattan Lal Sharma v. Managing Committee, (1993) 4 SCC 10: a member of an inquiry committee gave evidence against the employee and then continued as judge of his own testimony. Apprehension became “real.”
Official / subject-matter bias
The decision-maker is so identified with a policy or department that he cannot hear a challenge to it with an open mind. The Court is slower here. Mere employment by the State does not disable a civil servant from an inquiry. What disables is a closed official stake in the very question to be decided. Mineral Development Ltd. v. State of Bihar, AIR 1960 SC 468, is an early subject-matter discussion.
Judicial obstinacy / closed mind
State of West Bengal v. Shivananda Pathak, (1998) 5 SCC 513, defined judicial bias as a preconceived opinion or predetermination so fixed that the mind is not open to conviction. A prior legal view on a point of law, by itself, is not that condition. A judge who has already announced how this dispute will end, or who treats dissent as insult, may cross the line. Ranjit Thakur is the military illustration of a closed mind after earlier hostility.
Institutional and appearance bias
The Court has been cautious about treating a whole forum as biased. It will not say “every consumer commission is biased” as a rule of law. It will say that this member, this prior association, this pecuniary link vitiates this proceeding. That is why a cultural tilt in a special court — “hamari mahila” — is a problem of judicial ethics and public confidence, but it becomes a vitiating bias in Supreme Court doctrine only when it can be tied to a person, an interest, or a reasonable observer’s fear in the particular case.
3. Recusal: duty, discretion, and the fear of bench-hunting
India has no statute that lists when a Supreme Court or High Court judge must recuse. Practice is conscience plus precedent.
SCAORA v. Union of India (NJAC, 2015–16) is the fullest recusal debate at Constitution Bench level. Recusal of Justice J.S. Khehar was sought because he sat in the Collegium that NJAC would replace. The plea failed. The Bench drew three working pointers that later High Courts still recite:
- Pecuniary interest — automatic exit.
- Interest in a cause promoted by a party (the Pinochet analogy) — automatic or near-automatic exit.
- Other links — apply the real-danger / reasonable-apprehension test; do not recuse on imaginary or tactical allegations.
Justice Kurian Joseph, in a concurrence, pressed that reasons for recusal should be indicated in the interest of transparency. That has not become a binding rule. In 2024 the Court told a petitioner that an inquiry into why a judge recused is unnecessary and harmful to independence (Chandraprabha withdrawal matter). Recusal remains largely unreasoned and unreviewable.
Subrata Roy Sahara v. Union of India, (2014) 8 SCC 470, is the anti-forum-shopping judgment. Recusal sought of judges who had already held Sahara in contempt was rejected. The Court said the path of recusal is often “a convenient and a soft option,” and a litigant cannot be allowed to browbeat the court into a chosen bench.
Neelam Manmohan Attavar v. Manmohan Attavar (2021): a previous adverse order by the same judge is not, by itself, a ground to demand recusal.
Indore Development Authority v. Manoharlal (2019–20): Justice Arun Mishra refused to recuse from a larger bench on land acquisition merely because he had decided the point in a smaller combination. A prior legal holding is not personal stake. Whether that refusal satisfied the *Ranjit Thakur * observer is still argued in the academy; as precedent, the order stands for the proposition that disagreement with a judge’s law is not bias.
Recent restatement: Singhania University line (2025), Justices Gavai and Masih — “justice should not only be done but also seen to be done”; if a party has a genuine apprehension of conflict (for example the judge earlier appeared as standing counsel on a similar issue), courts should normally recuse. That is hortatory. The same Court still refuses recusal when the earlier matter is distinct and the plea looks like bench hunting.
Chandra Kumar Chopra v. Union of India (2012): apprehension must match the prudence of a reasonable person, not wild or irrelevant allegation meant to stall the case.
4. What does not count as bias
The Court has repeatedly refused to collapse these into disqualification:
- A judge belonging to a community, gender, or social group that the case touches.
- A judge having written extra-judicially on a legal topic, unless the writing prejudges this lis.
- An adverse interlocutory order.
- The mere fact that the State is a party and the judge is a State functionary.
- A special court created by Parliament for a class of complainants. Creation of mahila, POCSO, or SC/ST special courts is legislative policy under Articles 14/15(4), not a Kraipak defect. Bias has to be shown in the holder of the office, not in the existence of the office.
Waiver remains live. Manak Lal still bites: sit through the inquiry knowing the member was your old opponent, and you will not be heard to scream bias after you lose.
5. How the precedents fit together
SituationLeading caseResultMember of board is himself a candidateKraipak (1969)Real likelihood; selection quashedTribunal member earlier counsel against the partyManak Lal (1957)Appearance of bias; but can be waivedConfirming officer already hostileRanjit Thakur (1987)Reasonable apprehension; look through the litigant’s eyesInquiry member turns witness then judgeRattan Lal Sharma (1993)Real biasAuthor selects his own bookMohapatra (1984)Likelihood of biasDirect money interestNJAC recusal discussion; practice since Manak LalAutomatic disqualificationJudge earlier decided the same point of lawIndore Development Authority; Shivananda PathakGenerally not recusalJudge earlier passed adverse orders / contemptSahara (2014); Attavar (2021)Recusal refused; no bench huntingCollegium member hearing NJAC validitySCAORA (2015)Recusal refusedDemand to probe why a judge recused2024 SC oral orderInquiry rejected
6. The gap the reports do not close
Doctrine is clear at the level of the individual adjudicator. It is thinner at the level of institutional atmosphere — the unrecorded prior that a designated court exists for one class of party. Supreme Court bias law can invalidate a selection board that contains a rival. It has not developed a separate test for “this forum’s official motto is protection of X, therefore every accused is denied an open mind.” That claim still has to be translated into Ranjit Thakur language: would a reasonable person, knowing how this bench spoke and ruled, fear that the hearing was not fair?
That translation is possible in a particular case (a chair who announces that “our woman must not be harmed,” then shuts down cross-examination). It is not how the Court strikes down a statute creating special courts. Union of India v. State of Maharashtra (2019), on the SC/ST Act, treated protective discrimination as constitutional policy, not as a bias defect in the special court as such.
So the precedent map is this:
- Personal, pecuniary, and role-conflict bias — strict, old, and still expanding from courts into administration.
- Judicial recusal — conscience-based, anti-tactical, light on reasons, heavy on “don’t hunt the bench.”
- Forum-level or ideological tilt — a political and ethical problem the Court recognises in passing (Arnesh Kumar on arrest culture; Jacob Mathew on treating every doctor as a criminal) but has not recast as a free-standing ground to dissolve a special jurisdiction.
Anyone arguing bias in the Supreme Court today still wins or loses on Kraipak and Ranjit Thakur: not “I lost, therefore the court was against my kind of person,” but “a fair-minded observer, knowing these concrete facts about this decision-maker, would think the scales were not level.”
deemagclinic
deemag and chest clinic 