The 1999 Vajpayee victory is usually narrated as the moment the Bharatiya Janata Party came of age: a National Democratic Alliance total of 303 Lok Sabha seats, a government sworn in on 13 October 1999, and a prime minister who carried his administration to the end of its effective term. That narration is accurate as chronicle and wrong as analysis. Read against the Election Commission of India’s statistical report on the 1999 general election, the verdict demonstrates something narrower and more consequential. The BJP won exactly the 182 seats it had won in 1998, on a smaller share of the vote, and the expansion in the alliance total came from partners the party had acquired rather than from electors it had converted. What collapsed in 1999 was the Indian National Congress, which fell to 114 seats, the lowest count the party had recorded since independence. The victory was substantively a Congress collapse, and the failure to read it that way, by the victors above all, set up the strategic miscalculation that ended in the India Shining campaign and the shock verdict of 2004.

The Standard Reading of the 1999 Vajpayee Victory: Stability at Last

The celebratory reading rests on genuine achievements, and it deserves to be stated at its strongest before it is tested. Three general elections had been held in the four years from 1996 to 1999, and four governments had occupied office in New Delhi across that span: the thirteen-day Vajpayee ministry of May 1996, the United Front administrations of H. D. Deve Gowda and I. K. Gujral sustained by Congress backing from outside, and the second Vajpayee government of March 1998, which fell on 17 April 1999 when it lost a confidence vote in the Lok Sabha by 269 votes to 270 after the All India Anna Dravida Munnetra Kazhagam withdrew its backing. Against that record of fragility, the verdict counted on 6 October 1999 looked like deliverance. The Election Commission’s statistical report recorded 303 seats for the BJP and its pre-poll allies, and Atal Bihari Vajpayee took the oath as prime minister for the third time on 13 October 1999 at the head of an alliance whose majority did not depend on any single partner’s weekly goodwill.

The standard reading drew a second conclusion from the first: that the BJP had crossed a threshold of national acceptance. The party’s own deliberations after the verdict, recorded in the resolutions of its National Council in 1999, treated the result as a mandate for the BJP’s programme moderated through the National Agenda for Governance, and as proof that the coalition method had matured into a durable vehicle for the party’s growth. Biographical writing on the Vajpayee years, including Inder Malhotra’s profiles of the prime minister, reinforced the frame of a leader whose personal stature had finally been matched by his party’s electoral weight. The reading travelled because it flattered a real transition: for the first time, a non-Congress formation held a Lok Sabha majority with a reasonable expectation of keeping it.

Did the BJP win more votes than the Congress in 1999?

No. The Election Commission’s statistical report on the 1999 general election records the Congress polling 103,120,330 votes, a 28.30 percent share, against the BJP’s 86,562,209 votes and 23.75 percent. The Congress won the larger vote and the far smaller seat count, finishing with 114 seats to the BJP’s 182.

The inversion arose from distribution. Congress votes were spread across constituencies the party contested largely alone, whereas the BJP’s vote was concentrated inside alliance arrangements that converted regional partners’ strength into seats, so the larger national vote produced the smaller parliamentary party.

That single comparison, votes against seats, is the loose thread in the celebratory fabric. A party whose national vote share fell between 1998 and 1999, from 25.59 percent to 23.75 percent on the Election Commission’s figures, and whose seat count did not move at all, had not expanded. Something else had moved. The sections that follow identify the moving part: the Congress’s organisational collapse, the alliance arithmetic that disguised it, the Kargil attribution that decorated it, and the strategic lesson that the victors drew from a verdict they had misread. The frame used throughout is the one this series applies to the whole period, the coalition-era renegotiation framework, in which no national verdict after 1989 can be read off the largest party’s headline number alone, because the number is always a product of bargains struck state by state.

The evidence discipline matters here. Seat counts and vote shares in this article are the Election Commission of India’s, drawn from its statistical reports on the general elections of 1998 and 1999. Interpretations are anchored to named readers of the verdict: Yogendra Yadav and Suhas Palshikar, whose analyses rested on the National Election Study post-poll data gathered by the Centre for the Study of Developing Societies; Christophe Jaffrelot, whose essays on the BJP’s coalition compulsions appeared in the years immediately after the verdict; and Inder Malhotra, whose biographical writing on Vajpayee represents the strongest form of the leader-centred reading this article argues against. Where these readers disagree, the disagreement is the subject, and it is adjudicated on the numbers.

The Numbers Underneath: A Congress Seat Count of 114

Begin with the count that the standard reading hurries past. The Election Commission’s statistical report on the 1999 general election gives the Congress 114 seats. No Congress result since the first general election of 1951 to 1952 had been that low. The party’s previous floor was the 154 seats of 1977, won in the verdict that punished the Emergency, and even that defeated Congress of 1977 had won 40 more seats than the Congress of 1999 managed in a year without an Emergency to answer for. The fall from the immediately preceding election was steep in its own terms: 141 seats in 1998 became 114 in 1999, a loss of 27 seats in eighteen months, on a vote share that had actually risen from 25.82 percent to 28.30 percent. A party gaining votes and haemorrhaging seats is a party whose machinery for converting votes into victories has broken, and the breakage, not the victor’s strength, is the event of 1999.

The comparison that settles the argument is the paired reading of the two elections, and it is set out in the table that follows this paragraph, the 1998 to 1999 Balance Table, compiled from the Election Commission’s statistical reports for the two elections. Readers who work with the commission’s published volumes in digital form, turning their constituency tables into spreadsheets for state-level comparison, commonly open them through an office file viewer for spreadsheet and document formats; the figures below are the report-level aggregates on which the article’s claim stands or falls.

Measure (Election Commission statistical reports)	1998	1999	Movement
BJP seats	182	182	None
BJP vote share	25.59 percent	23.75 percent	Down 1.84 points
Congress seats	141	114	Down 27
Congress vote share	25.82 percent	28.30 percent	Up 2.48 points
Seats won by BJP pre-poll allies	70	121	Up 51
BJP-led alliance total	252	303	Up 51
Congress and pre-poll allies, combined seats	Above 160	136	Sharply down

Every row of the 1998 to 1999 Balance Table points the same way. The BJP’s seat line is flat and its vote line is negative. The alliance’s growth line is identical, seat for seat, to the allies’ growth line: the entire expansion of the BJP-led total between 1998 and 1999 is accounted for by partner acquisition. The Telugu Desam Party won 29 seats in Andhra Pradesh in 1999 inside the alliance arrangement, the Janata Dal (United) won 21 in Bihar and Karnataka, the Dravida Munnetra Kazhagam won 12 in Tamil Nadu, and the Biju Janata Dal won 10 in Orissa, each of them a regional formation whose votes the BJP did not command and could not transfer. Meanwhile the Congress’s seat line fell by a fifth in a single cycle even as its vote share recovered by two and a half points. The verdict’s arithmetic is not a story of one party rising. It is a story of one party’s coalition expanding across the space vacated by another party’s collapse.

Which state gave the alliance its largest allied contingent in 1999?

Andhra Pradesh. The Election Commission’s statistical report records the Telugu Desam Party winning 29 of the state’s 42 Lok Sabha seats in 1999, the largest single allied contingent in the National Democratic Alliance, ahead of the Janata Dal (United) with 21.

The Telugu Desam tally exceeded the BJP’s own winnings in every state that year, and it was delivered by Chandrababu Naidu’s state organisation and his government’s record in Andhra Pradesh, not by any BJP machinery on the ground.

The Congress’s seat collapse had a geography, and the geography was organisational rather than ideological. In Uttar Pradesh, the state’s 85 seats returned 29 BJP members in 1999, a fall from the 57 the party had won there in 1998, and the Congress, which had once treated the state as its citadel, won 10. In Bihar, the Rashtriya Janata Dal and the Congress between them could not hold the ground that the Janata Dal (United) and BJP arrangement took. In West Bengal, the Trinamool Congress’s 8 seats came substantially out of the Congress’s old electorate, and the Congress was reduced to a remnant in a state it had governed for most of the period from 1972 to 1977 and contested seriously for decades after. In Tamil Nadu, the Congress fought in the AIADMK’s company and won nothing like a revival. State after state, the pattern repeats: the Congress’s votes survived in fragments, its seat-winning apparatus did not.

The point about vote share deserves emphasis because it disciplines the collapse claim against exaggeration. The Congress of 1999 was not repudiated by the electorate in the manner of 1977; its 28.30 percent was the largest vote share won by any single party that year, larger than the BJP’s. What failed was everything that turns a national vote into a parliamentary party: pre-poll alliances that pool votes instead of splitting them, state units capable of selecting candidates with local standing, and a campaign organisation that concentrates effort where seats are winnable. The Congress contested the 1999 election substantially alone, its few arrangements local and late, and paid for solitary pride with the worst seat count in its history. Collapse, in this article’s usage, is an organisational diagnosis with a numerical signature, and the signature is the 27-seat fall on a rising vote.

An Organisation Allowed to Decay: The Congress Between 1991 and 1999

The collapse had a history, and the history begins with the paradox of 1991. The Congress government of P. V. Narasimha Rao carried through the economic reform programme whose political management is traced in the account of the 1991 reform’s institutional consummation, yet the same five years saw the party’s organisation hollowed out at every level below the cabinet room. Rao held the Congress presidency together with the prime ministership after Rajiv Gandhi’s assassination in May 1991, and the dual charge reproduced an older deformation: the party existed to sustain a government, and the government had no time to rebuild the party. Organisational elections inside the Congress, never robust after the early 1970s, lapsed into nomination at the state level. Pradesh Congress Committees became arenas for factional managers whose authority ran upward to Delhi rather than downward to districts. The district and block committees, which in the party’s functioning decades had identified candidates, managed booths and disciplined local notables, atrophied into letterheads.

The inheritance behind that decay reaches further back, to the reconstruction of the party around a single office after 1980, the post-1980 Congress(I) organisational-architecture inheritance under which internal elections gave way to nomination and the high command’s writ replaced the provincial bosses’ bargaining. A party rebuilt as an instrument of one leader’s authority can win landslides for as long as the leader’s authority is unchallenged, as the Congress did in 1984. The same design fails silently when authority is contested or absent, because no intermediate layer holds independent legitimacy. The terminal crisis of that design’s founding decade, examined in the account of the office-rebuilding’s legible end as Congress-decline antecedent, left the party entering the 1990s without the provincial leadership stratum that had once absorbed shocks. What Rao’s five years added was not recovery but drift at the top of a structure already missing its middle.

Sitaram Kesri’s presidency, from September 1996 to March 1998, completed the sequence. Kesri had been the party’s treasurer and a fixture of its back rooms; elevated after Rao’s resignation in the aftermath of the 1996 defeat, he presided over the Congress’s support, given from outside and withdrawn twice, to the United Front governments. The political consequence was corrosive in a way the seat numbers alone would later confirm. Between 1996 and 1998 the Congress was close enough to power to share its discredit and too far from office to claim its achievements, and its state units spent those years litigating factional claims in Delhi instead of contesting the BJP’s expanding state organisations. Kesri’s exchanges with Prime Minister H. D. Deve Gowda, culminating in the withdrawal of backing in March 1997 and Gowda’s replacement by I. K. Gujral, were conducted as palace transactions; the party’s district machinery was neither consulted nor rebuilt by any of them.

The deeper structural point is the one this series has traced from the founding: the Congress System that Rajni Kothari described in Politics in India, published in 1970, had rested on the party’s internal pluralism, its capacity to house factions that bargained inside one organisation rather than splitting into rival parties. The rupture that began with the 1969 split’s structural-cultural rupture converted that internal bargaining into external fragmentation over three decades. By the 1990s, the fragmentation had a name and an address in almost every state: former Congress formations, from the Trinamool Congress in West Bengal to the breakaway parties of the northeast and the Janata fragments of the Hindi belt, competed for the Congress’s own historical electorate. Each split was small. Their sum, visible in the 1999 constituency returns, was a national party bleeding votes to its own descendants in precisely the seats where a few thousand votes decided the result.

The organisational diagnosis explains a feature of the 1999 verdict that the strength reading cannot: the Congress’s losses were worst where its machinery had decayed furthest, not where the BJP’s appeal was newest. In Madhya Pradesh and Rajasthan, where the Congress had won assembly elections in December 1998 under Digvijaya Singh and Ashok Gehlot, its Lok Sabha performance in 1999 held up measurably better than in Uttar Pradesh or Bihar, where no functioning state organisation had existed for a decade. Electorates had not sorted themselves by enthusiasm for the BJP. They had sorted themselves, state by state, by whether a Congress organisation still existed to convert sympathy into seats.

What 28 Percent Used to Buy: The Congress’s Vote-Seat Arithmetic Across Decades

The organisational reading of the Congress’s 114 can be tested against the party’s own history, because the Congress has polled comparable vote shares before and converted them at wildly different rates, and the variable that explains the conversion rate is never the vote itself. In 1977, punished for the Emergency by an opposition united into a single Janata Party across the north, the Congress polled 34.52 percent of the national vote on the Election Commission’s figures and won 154 seats: a higher vote than 1999’s, converted under conditions of maximum opposition unity into the party’s previous record low. In 1989, the Congress polled 39.53 percent, the largest vote share of that election by a wide margin, and won 197 seats, losing office because the opposition’s seat-level arrangements in the north and south pooled the anti-Congress vote behind single candidates in state after state. In 1984, the party polled 48.1 percent and won 404 seats, the landslide that closed the founding era’s long story. Set the four results in a row, 404 seats on 48.1 percent, 197 seats on 39.53 percent, 154 seats on 34.52 percent, and 114 seats on 28.30 percent, and the pattern instructs: the Congress’s seat harvests track its vote only through the mediation of two organisational variables, the unity or division of the opposition arrayed against it in each state, and the presence or absence of the Congress’s own arrangements and machinery to defend its vote constituency by constituency.

The 1999 figure is the extreme of the series because both variables stood at their worst for the Congress simultaneously. Opposition unity, in the form of the National Democratic Alliance’s seat divisions, reached its post-1989 maximum: in the great majority of constituencies the Congress faced a single alliance candidate rather than the divided fields of 1996. The Congress’s own machinery and arrangements stood at their post-1947 minimum, for the reasons the preceding sections detailed. The vote share, 28.30 percent, was the residue: the loyalty of the party’s remaining electorate, delivered without conversion. Political commentary that treats vote share as the measure of a party’s health will read 1999 as a Congress holding its own, since 28.30 percent exceeded the BJP’s share and improved on 1998. Commentary that reads seat conversion as the health measure will see the truth the party’s own Antony Committee saw: a party can retain the affection of more than a quarter of the electorate and still be dying as a parliamentary organisation, because parliamentary organisations live on conversion, and conversion is machinery. The 1999 Congress was not rejected. It was unconverted, and the distinction is the difference between a party that needs a better argument and a party that needs to be rebuilt. The rebuilding, as the later sections show, is what the party proceeded to do, while the victors, reading their own conversion of allied votes as affection for themselves, rebuilt nothing.

The Simultaneous Verdicts: What the State Assemblies Said in October 1999

The 1999 general election was held together with assembly elections in several states, and the simultaneous verdicts supply a control experiment that the national reading rarely uses. The same electorates, voting on the same days, rewarded the Congress where the Congress possessed a functioning state organisation and punished it where it did not, while their Lok Sabha choices followed the alliance arithmetic of their states. The variable that explains the difference is organisation, and the experiment is as close as electoral history comes to isolating it.

Karnataka is the cleanest case. The Congress fought the October 1999 assembly election under S. M. Krishna, a state leader with an intact organisation and a campaign built on the failures of the incumbent Janata Dal (Secular) and Janata Dal (United) formations, and it won the state, forming the government in Bangalore. Its Lok Sabha performance in Karnataka rose with the state machinery’s effort, and Sonia Gandhi’s own victory at Bellary sat inside a state-level Congress effort that was actually organised. Andhra Pradesh supplies the counter-case in the same week. The Congress’s vote share in the state rose against the Telugu Desam government, yet the Telugu Desam retained the assembly and swept the Lok Sabha seats in the pooling arrangement with the BJP, because the Congress’s state organisation, divided between rival claimants for the chief ministership, could not convert a competitive vote into constituency victories. Maharashtra supplies the third case: the Congress and the Nationalist Congress Party, having split in May, together polled a vote that would have dominated the state, and together formed the government afterward in an arrangement of necessity, yet their division through the campaign handed the BJP and Shiv Sena alliance the larger Lok Sabha harvest. In each state the Congress’s problem was the same deficiency wearing a local costume: leadership unsettled, organisation factionalised, arrangements improvised or refused. In the one state where the deficiency had been repaired in time, the party won. The national tally of 114 seats aggregates these state stories; it does not contradict them, and no appeal to the BJP’s national magnetism can explain why the magnetism failed precisely where the Congress had repaired itself.

Two Hundred and Seventy-Two: The Dress Rehearsal of April 1999

The collapse was previewed in April 1999, in the eleven days between the fall of the second Vajpayee government and the dissolution of the twelfth Lok Sabha, and the preview deserves attention because it isolates the Congress’s deficiency in its purest form. When the confidence vote went against the government on 17 April 1999, President K. R. Narayanan followed constitutional convention and invited the leader of the largest opposition party to attempt a government. Sonia Gandhi met the President on 21 April 1999 and told the press, in a sentence that followed her for years, that the Congress had 272 members and more were joining. The arithmetic was the claim of a working majority. The arithmetic was also false, and its falsity was demonstrated within days by the refusals of the very leaders whose support the count presumed.

Mulayam Singh Yadav, whose Samajwadi Party held 20 seats, declined to back a Congress-led government and made his refusal public, insisting on terms the Congress would not meet. J. Jayalalithaa, whose withdrawal had created the vacancy, proposed Jyoti Basu of the Communist Party of India (Marxist) as prime minister instead, and the Left’s own leadership declined the honour its politburo would not authorise Basu to accept. The Congress, which had spent two years withdrawing and restoring its backing to United Front governments as an exercise of leverage, discovered in April 1999 that leverage accumulated over years of palace transaction could not be converted into a coalition on demand. On 25 April 1999, Sonia Gandhi informed the President that she could not form a government, and the Lok Sabha was dissolved the following day. The party went into the general election having publicly demonstrated, to every regional leader in the country, that it could neither assemble a majority nor be relied upon to sustain someone else’s.

The April failure matters to the reading of the autumn verdict for two reasons. The first is mechanical: the regional formations that the Congress would need as partners in any future arrangement priced the Congress’s weakness into their decisions through the summer of 1999, and most of them declined to deal with it at all. The National Democratic Alliance’s ledger of partners, examined in a later section, grew through precisely the months in which the Congress sat without a single major pre-poll arrangement outside a handful of states. The second reason is interpretive. The strength reading of 1999 requires the election to have been a contest between a rising party and a declining one, decided by the electorate’s preference. The April record shows the contest substantially settled before the electorate was consulted: the Congress entered the campaign having failed the coalition test in public, and the verdict’s seat distribution ratified a judgement the political class had already delivered. Electorates in 1999 chose governments constituency by constituency. The menu from which they chose had been written by the Congress’s organisational and diplomatic incapacity in April.

A Succession Unsettled: Sonia Gandhi’s First Campaign as President

Into this hollowed structure stepped a president who had held the office for eighteen months when the 1999 election was called. Sonia Gandhi became Congress president on 14 March 1998, when the Congress Working Committee removed Sitaram Kesri and installed her in his place, an accession by resolution rather than by organisational election. Her authority over the party’s old guard was real but untested, her network among state leaders thin, and the machinery she inherited was the one Kesri’s presidency had left behind. The 1999 campaign was consequently two campaigns at once: a national appeal built around her candidacy for the prime ministership, and a state-level scramble in which candidate selection, alliance negotiation and campaign logistics were improvised constituency by constituency.

The fragility of the succession was exposed in May 1999, weeks after the fall of the Vajpayee government and before the election calendar was set. On 15 May 1999, Sharad Pawar, P. A. Sangma and Tariq Anwar wrote to Sonia Gandhi arguing that the Congress should project a prime minister born in India, a public challenge to her eligibility framed around her Italian birth. The Congress Working Committee expelled all three on 20 May 1999, and the expelled leaders founded the Nationalist Congress Party on 25 May 1999. The episode cost the Congress in Maharashtra, where Pawar’s network was the party’s surviving organisation in the state’s western districts, and the Nationalist Congress Party’s 8 seats in the 1999 returns were won substantially on ground the undivided Congress had held. A succession that produces a splinter party in its first electoral season has not consolidated, whatever the personal popularity of the leader at its head.

Where did Sonia Gandhi contest the 1999 election from?

From two constituencies in two states. Sonia Gandhi stood at Amethi in Uttar Pradesh, the seat associated with Rajiv Gandhi, and at Bellary in Karnataka, and the Election Commission’s statistical report records her winning both, at Amethi by a margin above three lakh votes.

She retained Amethi and vacated Bellary, and the twin victories demonstrated personal vote-pulling power on a scale no other Congress candidate matched in that election. The party’s 114 seats demonstrated the limit of that power: it had not yet been converted into organisation.

The contrast inside that answer is the section’s argument in miniature. Sonia Gandhi’s own contests were triumphs of candidate-centred campaigning: the Amethi margin, and the Bellary victory in a state where the Congress was simultaneously winning an assembly election under S. M. Krishna, showed an electorate willing to vote for her. The party beneath her won 114 seats. Between the leader’s margins and the party’s tally lay the missing middle of the Congress, the state and district machinery whose absence no leader’s popularity could paper over in 543 separate constituency contests. The organisational rebuilding that Sonia Gandhi’s presidency eventually undertook belongs to the period after 1999; the election itself caught the succession at its most exposed point, after the old order had been dismissed and before a new one had been constructed.

The BJP understood the exposure and campaigned on it, making the Congress president’s foreign birth a national theme through the summer of 1999. The theme’s electoral weight is contested among analysts of the verdict, and this article treats it as one element of the campaign environment rather than a cause of the seat count. Votes lost on a question of the leader’s origin would appear in the Congress’s vote share; the share rose. What the origin campaign did achieve was to keep the Congress’s campaign defensive and leader-focused through the very months in which alliance negotiations and candidate selection demanded the party’s full organisational attention. The succession question and the organisation question were the same question, and the Congress went into the election with both unanswered.

One Hundred and Eighty-Two, Twice: The BJP’s Frozen Seat Count

Turn from the loser to the winner, and the winner’s numbers refuse to perform the growth the standard reading assigns them. The BJP won 182 Lok Sabha seats in the general election of February and March 1998. It won 182 Lok Sabha seats in the general election of September and October 1999. Between the two victories, the party contested as the senior partner of a broader alliance, campaigned behind a prime minister whose personal ratings exceeded his party’s in every published survey of the period, and fought an election in the immediate aftermath of a victorious war. Its seat count did not move by one. Its vote share moved down, from 25.59 percent to 23.75 percent on the Election Commission’s statistical reports. Electoral peaks do not present themselves this way. A party at a rising peak converts favourable conditions into additional seats in additional states. The BJP of 1999 converted favourable conditions into a larger coalition and the same parliamentary party.

The state detail sharpens the point. In Uttar Pradesh, the BJP’s largest hunting ground, the party fell from 57 seats in 1998 to 29 in 1999, a collapse inside the victory that the national total conceals. The losses in the Hindi belt were offset in the alliance ledger by partner gains elsewhere: the Janata Dal (United), assembled in 1999 from the Samata Party, the Janata Dal fragments of Bihar and Karnataka and the Lok Shakti, won 21 seats where its components had won a scattering before; the Dravida Munnetra Kazhagam, contesting in the BJP’s company for the first time, won 12 seats in Tamil Nadu, a state where the BJP had no independent presence worth the name; the Telugu Desam Party’s 29 seats in Andhra Pradesh were won under Chandrababu Naidu’s leadership on his government’s record, in an arrangement completed late in the campaign. The alliance that took office in October 1999 was broader than the one that had fallen in April. The party at its centre was, by the only measures elections supply, the same size.

The footprint evidence completes the picture. The Election Commission’s statistical reports record the BJP contesting 388 seats in 1998 and 339 seats in 1999: the party stood aside in 49 additional constituencies to accommodate the partners its majority required. Its strike rate, the proportion of contested seats won, rose from 47 percent to 54 percent, and party publicists used the strike rate after the verdict as evidence of sharpened strength. The inference inverts the causality. A party that withdraws from the constituencies it cannot win will always post a finer strike rate, and the withdrawal itself is the datum: the BJP’s national footprint, the territory in which it could present its own candidate with a serious prospect, had contracted between its two victories. In the states where it did expand its candidature, its returns did not improve. Growth that requires the grower’s absence from the field is a property of the alliance’s design, and the design’s terms were dictated by the partners’ strength in their own territories, not by the BJP’s advance into new ones.

What did the BJP’s National Council claim the 1999 verdict meant?

The BJP’s National Council resolutions of 1999 read the verdict as a mandate for stability under Vajpayee’s leadership and as an endorsement of governance conducted through the National Agenda for Governance, the alliance’s common programme, from which the party’s three core commitments had been deferred to secure allied agreement.

The three commitments were a Ram temple at Ayodhya, the abrogation of Article 370 and a uniform civil code. The resolutions treated the alliance’s parliamentary majority as the party’s own arrival at durable national power, the interpretation this article disputes on the Election Commission’s figures.

The mechanics of the expansion had been rehearsed the year before, in the 1998 NDA formation’s strategic-paradox context, where the BJP first accepted that its route to office ran through partners whose price was the deferral of the party’s defining commitments. What changed in 1999 was the quality of the merchandise on offer to partners. A caretaker prime minister who had conducted a war to a successful conclusion between May and July 1999, and who had lost office by a single vote in April, was a more attractive alliance anchor than the leader of 1998 had been. Partners were not joining a rising party. They were joining a plausible government, and the distinction governed everything that followed, because a plausible government’s partners remain partners only for as long as the plausibility lasts, and they draw their own conclusions when the anchor’s party begins to mistake borrowed seats for owned ones.

Christophe Jaffrelot’s essays on the BJP’s coalition compulsions, collected in the volume The BJP and the Compulsions of Politics in India, published in 2000 with Thomas Blom Hansen, made this argument in the verdict’s immediate aftermath. Jaffrelot read the BJP’s expansion across the 1990s as real but bounded: the party’s social coalition had widened, its geographic reach had extended into the south and east through alliances, and its core electorate had not grown in proportion to its parliamentary presence. The 1999 numbers fit his reading precisely. The party’s seat count in its strongholds stagnated or fell, its vote share slipped, and the parliamentary majority materialised in the accounts of its partners. The strength reading requires the BJP to have grown. The record shows an alliance that grew around a party that had not.

The Alliance Ledger: How the 121 Partner Seats Were Assembled

The allied growth from 70 seats in 1998 to 121 in 1999 was not a tide. It was a ledger, assembled entry by entry through negotiations whose terms reveal how little of the total the BJP itself supplied. The largest entry was Andhra Pradesh. Chandrababu Naidu’s Telugu Desam Party had sustained the United Front experiment, then held aloof from both national formations, and its arrangement with the BJP for 1999 was completed late, on a seat division that left the Telugu Desam contesting the great majority of the state’s 42 seats. The Election Commission’s statistical report records the outcome: Telugu Desam 29 seats, BJP 7 in the state. The alliance’s Andhra harvest of 36 seats was gathered by Naidu’s organisation, on the record of Naidu’s state government, and the Congress’s strong vote in the state, higher than its 1998 share there, was defeated by the pooling arrangement rather than by any BJP presence.

The second entry was Bihar, and it was an act of party construction. The Janata Dal, which had disintegrated as a national party after 1997, reassembled its anti-Lalu Prasad fragments in 1999 as the Janata Dal (United): the Samata Party of George Fernandes and Nitish Kumar, the Janata Dal factions that had broken with Lalu’s Rashtriya Janata Dal, and Ramakrishna Hegde’s Lok Shakti from Karnataka. The Election Commission’s report credits the new formation with 21 seats. Those seats had been won against the Rashtriya Janata Dal’s Bihar organisation in a contest whose terms were set by Bihar’s politics, and their contribution to the Delhi total was the conversion of the state’s anti-incumbency into alliance seats. The BJP’s own Bihar contingent formed the smaller part of the state’s alliance harvest.

The third entry was Tamil Nadu, and it was a substitution, the clearest single illustration in the ledger of what alliance breadth meant. In 1998 the BJP’s Tamil partner had been the AIADMK, whose 18 seats had made J. Jayalalithaa the alliance’s most demanding member and whose withdrawal in April 1999 destroyed the second Vajpayee government. In the 1999 election the AIADMK fought on the other side, in arrangement with the Congress. The BJP’s Tamil partner in 1999 was the AIADMK’s rival, the Dravida Munnetra Kazhagam of M. Karunanidhi, whose government in Chennai Jayalalithaa had demanded be dismissed as the price of her continued backing in Delhi. The Election Commission’s report records the DMK winning 12 seats and its smaller Tamil allies adding further seats to the alliance column, in a state where the BJP itself won 4. A formation that can exchange its principal partner in a major state for that partner’s bitterest rival, within a single electoral cycle, and emerge with its majority intact, is not displaying partisan depth. It is displaying brokerage, the capacity of a plausible prime minister to be the counter through which regional ambitions are settled. Brokerage is a governing skill. It is not electoral strength, and it evaporates on the day the counter’s plausibility does.

The remaining entries repeat the pattern at smaller scale. In Orissa, the Biju Janata Dal of Naveen Patnaik won 10 seats on the strength of a state organisation built around Biju Patnaik’s legacy, in a formal seat division with the BJP. In Maharashtra, the Shiv Sena’s 15 seats, up from 6 in 1998, were won by Bal Thackeray’s organisation in the Sena’s own strongholds, in a state where the Congress’s vote was simultaneously divided by the Pawar split described in the preceding section. In West Bengal, Mamata Banerjee’s Trinamool Congress won 8 seats inside the alliance, seats carved from the Congress’s historical electorate by a leader who had left the Congress in 1997 precisely because its organisation could not accommodate her. In Haryana, Om Prakash Chautala’s Indian National Lok Dal added 5. Entry by entry, the ledger’s 121 allied seats resolve into regional organisations with their own leaders, their own electorates and their own reasons, renting their totals to a national anchor. The anchor’s contribution was Vajpayee’s plausibility and the absence, after the Congress’s April failure, of any competing counter. Neither contribution appears in the BJP’s own 182.

Architecture Before Strength: What the Coalition Total Actually Measured

The strongest objection to the Congress-collapse reading must be faced in its strongest form. The government formed on 13 October 1999 governed. It carried a full legislative programme across four and a half years: the Insurance Regulatory and Development Authority Act of 1999 opened the insurance sector; the Foreign Exchange Management Act of 1999 replaced the emergency-era exchange control regime; a Department of Disinvestment was created in December 1999 and elevated to a ministry; the National Highways Development Project, with the Golden Quadrilateral at its centre, moved from approval to construction; the Empowered Committee of State Finance Ministers was constituted in 2000 and the Kelkar task forces of 2002 laid fiscal groundwork on which the goods and services tax would eventually be built. Abroad, the government consolidated India’s standing as a declared nuclear-weapon state after the tests of May 1998, sustained the Jaswant Singh and Strobe Talbott dialogue with the United States through 2000, and hosted President Bill Clinton’s visit of March 2000. If the victory had been mere arithmetic over a collapsed opponent, the objection runs, the government it produced would not have been capable of this record.

The objection proves less than it claims. Every item in that record was made possible by the property the verdict actually conferred, which was durability of the coalition architecture rather than depth of the BJP’s support. A government whose majority rests on partners with state-level interests of their own can legislate energetically in the domains where partner interests align, economic administration above all, since Chandrababu Naidu’s Andhra Pradesh and the other allied state governments were themselves pursuing investment, infrastructure and fiscal credibility. The same architecture enforced the deferrals recorded in the National Agenda for Governance: the temple, Article 370 and the uniform civil code stayed deferred because the architecture’s price was their deferral. Durability, in other words, was a property of the bargain, and the bargain was available because the Congress’s collapse had left the alliance’s partners with no alternative national anchor. The Congress-collapse reading does not deny the government’s record. It explains the conditions that made the record possible, and it predicts, correctly, that the record would be misread by its authors as evidence of partisan growth.

There is a subtler confirmation in the government’s internal distribution of credit. The portfolios that carried the administration’s reputation, finance under Yashwant Sinha and then Jaswant Singh, disinvestment under Arun Shourie, external affairs, the prime minister’s own office, were BJP-held, and the party understandably experienced the government’s competence as its own. Experiencing coalition governance as party strength is the natural psychology of a senior partner. It is also the precise mechanism by which the misreading propagated from the verdict of 1999 into the strategy of 2004, since a party that believes its government’s record is its own electoral asset will campaign on the record’s glow rather than on the constituency arithmetic that actually decides its fate. The 2004 campaign’s fate is the subject of the final sections; the mechanism belongs here, at the point where architecture was first mistaken for strength.

The Party Table Read Whole: Movement Beneath a Stable Headline

The Election Commission’s statistical report permits one further test that neither camp’s rhetoric usually survives: reading the complete party table of 1999 against that of 1998, formation by formation, and asking where the churn actually occurred. The headline parties present a study in opposites. The BJP’s line is motionless, 182 seats on a diminished vote. The Congress’s line is a fall of 27 seats on an enlarged vote. Beneath them, the table is in ferment. The Shiromani Akali Dal, the BJP’s Punjab partner, collapsed from 8 seats in 1998 to 2 in 1999, a decline the alliance’s national total absorbed without comment because other entries compensated. The Rashtriya Janata Dal fell from 17 to 7. The Communist Party of India fell from 9 seats to 4, continuing its long contraction outside the Left Front’s fortresses, while the Communist Party of India (Marxist) held at 33, its stability a product of state organisations in West Bengal, Kerala and Tripura as disciplined as any in the country. The Asom Gana Parishad, which had governed Assam within living memory, won nothing at all. The Samajwadi Party rose from 20 seats to 26 and the Bahujan Samaj Party from 5 to 14, the two Uttar Pradesh formations harvesting between them the ground the BJP and the Congress both vacated in the state where the largest single contingent of the Lok Sabha is elected.

Read whole, the table describes a system in which organisational health, state by state, predicted outcomes with a fidelity no national narrative matched. Parties with intact state machines, the Telugu Desam, the Dravida Munnetra Kazhagam, the Biju Janata Dal, the Left Front’s constituents, the Samajwadi Party, held or advanced. Parties whose machines had decayed, the Congress in most of the map, the Akali Dal in Punjab, the Rashtriya Janata Dal in Bihar, the Janata Dal’s remnants wherever the Janata Dal (United) had not reassembled them, fell regardless of their national leaders’ prominence or their place in the winning and losing columns. The 1999 verdict, disaggregated to the level at which Indian elections are actually fought, is an audit of party organisations conducted by the electorate. The BJP’s distinction was not that its organisation had passed the audit with honours; its Uttar Pradesh failure recorded the opposite. Its distinction was that its leader’s plausibility and its opponents’ disarray allowed it to purchase, at the price of portfolios and deferrals, the passing grades of twenty-odd other organisations and present the aggregate transcript as its own.

The Ministry as Ledger: What the Government’s Own Composition Admitted

The government sworn in on 13 October 1999 published the verdict’s true anatomy in its own composition. The council of ministers Vajpayee assembled was among the largest of the republic’s history to that date, and its size was a function of the ledger: portfolios had to be found for the Shiv Sena, the Janata Dal (United), the Biju Janata Dal, the Trinamool Congress, the Dravida Munnetra Kazhagam, the Pattali Makkal Katchi, the Indian National Lok Dal, the Jammu and Kashmir National Conference and a tail of single-member formations, in addition to the BJP’s own claimants. George Fernandes, the alliance’s convenor, took Defence; Nitish Kumar of the Janata Dal (United) took Railways; Mamata Banerjee took Railways’ counterpart prominence for the Trinamool in the same portfolio’s rotation across the term’s reshuffles; Sharad Yadav, Ram Vilas Paswan and Omar Abdullah carried allied colours in the cabinet room. The Telugu Desam Party, the ledger’s largest creditor, declined cabinet seats altogether and sustained the government from outside, taking instead the Speaker’s chair for G. M. C. Balayogi, whose election to preside over the Lok Sabha recorded the arrangement’s terms in the House’s own offices. A government whose largest supporting party prefers the Speaker’s chair and external leverage to cabinet responsibility is a government whose prime minister governs by continuous negotiation, and the negotiations never closed for the remainder of the term.

The composition carried a second admission. The portfolios the BJP reserved to itself, home affairs under L. K. Advani, finance, external affairs, the prime minister’s own charge of the atomic energy and space departments, were the portfolios of the state’s command. The portfolios the allies prized, railways, surface transport, agriculture, chemicals and fertilisers, civil aviation, were the portfolios of distribution, touching their state electorates directly. The division was not corruption of the cabinet principle; it was the coalition’s constitution, understood by every participant. It meant that the government’s daily operation ran on allied consent in the distributory domains where elections in the states are actually decided, and that the BJP’s freedom of action was greatest exactly where its 1999 mandate, such as it was, had least to say. The strength reading, applied to the ministry, would predict a senior partner dictating terms to supplicants. The ministry’s composition records supplication running in both directions, with the senior partner paying in the currency, distributory portfolios, that its own state organisations most needed to accumulate. Every reshuffle of the term, and there were several between 1999 and 2003, re-ran the same negotiation: the Trinamool’s departure in 2001 and return in 2003, the Pattali Makkal Katchi’s exit and the Dravida Munnetra Kazhagam’s eventual withdrawal in December 2003, the Indian National Lok Dal’s comings and goings, each repriced the majority. The government was stable. The stability was a subscription, renewed monthly, and the party that believed it owned the building because it held the largest room was misreading its tenancy agreement in the most literal sense available to politics.

The Third Force Ledger: The Parties That Were Neither

A verdict is also defined by the space neither principal occupied, and the 1999 returns map that space with unusual precision. The Election Commission’s statistical report records the Communist Party of India (Marxist) at 33 seats, its West Bengal and Kerala organisations intact and its national role, as in 1996 and 1998, that of a disciplined third force too small to govern and too organised to ignore. In Uttar Pradesh, the state’s 85 seats divided in a manner that humiliated both national parties’ pretensions: the Samajwadi Party won 26 and the Bahujan Samaj Party 14, so that Mulayam Singh Yadav and Mayawati between them commanded nearly half the state’s representation, and the Congress’s 10 seats placed it fourth in the state that had supplied its prime ministers for most of the period since 1947. In Bihar, Lalu Prasad’s Rashtriya Janata Dal fell from 17 seats in 1998 to 7, its losses harvested not by the BJP directly but by the Janata Dal (United) construction described earlier. The national parties, Congress and BJP together, polled barely half the national vote between them, 52.05 percent on the commission’s figures, and won 296 of 543 seats. India’s politics in 1999 was not a two-party system with a fringe. It was a system of state parties and state verdicts, temporarily sorted into two national columns by the alliance market, and the columns’ totals recorded the sorting rather than any national conversion.

The third force’s behaviour after the verdict confirms the market character of the sorting. The Telugu Desam Party sustained the government from within the alliance, guarding its Andhra interests throughout; the Left parties opposed the government while their state governments pursued investment policies convergent with the centre’s; Mulayam Singh Yadav, who had refused Sonia Gandhi in April, kept his party equidistant and his state ambitions central. None of these formations understood itself as a junior member of a BJP-led national movement, and their parliamentary conduct across 1999 to 2004, issue by issue, followed their state calculations with a fidelity that the alliance’s managers in New Delhi repeatedly found inconvenient. The 303 of October 1999 was a sum of such calculations. Treating the sum as a party’s strength was the error; the components never made the mistake themselves.

The Campaign: Bellary, Foreign Origins and the Stability Theme

The campaign itself, fought from early September to the beginning of October 1999, displayed the two parties’ respective conditions in concentrated form. The BJP’s campaign was the more professional production, managed at the centre with Pramod Mahajan among its principal organisers, and built on two themes: stability, the promise that a vote for the alliance ended the cycle of falling governments, and the person of Vajpayee, presented as the statesman who had won a war without losing his restraint. The Congress’s foreign-origin attack on Sonia Gandhi was pressed hardest in the campaign’s most symbolic single contest, at Bellary in Karnataka, where the BJP fielded Sushma Swaraj, among its most effective Hindi orators, against the Congress president in a constituency chosen to dramatise the question. Swaraj campaigned in Kannada-inflected outreach and lost by a margin in the tens of thousands, and the Bellary result entered the campaign’s folklore as evidence that the origin theme, whatever its resonance in the Hindi belt’s editorials, did not move the voters of a southern constituency against a candidate whose party organisation in the state was functioning. The episode is a small controlled experiment inside the larger one: where the Congress was organised, even the opposition’s most heavily promoted theme failed.

The Congress’s campaign, by contrast, was an improvisation around a person. Sonia Gandhi toured extensively and drew large crowds, and the party’s central message, that the Congress alone offered a national alternative, was undercut each week by the observable absence of the arrangements a national alternative requires. State-level negotiations with potential partners were concluded late, partially, or not at all; in Bihar the arrangement with the Rashtriya Janata Dal was struck on terms that satisfied neither organisation’s district workers; in Uttar Pradesh the party contested effectively alone against three entrenched competitors. Campaigns do not create organisations in five weeks. They reveal them, and the revelation of 1999, repeated across five phases of polling from 5 September to 3 October, was the disparity between a leader’s draw and a machine’s absence that the seat count then recorded.

The Man and the Party: Vajpayee’s Personal Vote

One component of the standard reading survives every numerical test, and honesty requires its weight to be recorded before the Kargil question is taken up. Atal Bihari Vajpayee was, in 1999, a personal electoral asset of a magnitude his party could not match. The National Election Study post-poll data analysed by the Lokniti team registered his lead over every rival in the electorate’s preference for the prime ministership by margins no Congress leader approached, and constituency reporting through the campaign recorded voters who intended to vote for the alliance’s regional candidate, or even described themselves as habitual Congress supporters, naming Vajpayee as their reason. Inder Malhotra’s biographical writing on Vajpayee built its reading of the era on exactly this foundation: a leader of parliamentary vintage stretching back to 1957, acceptable to social constituencies his party could not reach, whose personal moderation functioned as the alliance’s collective letter of credit. On the evidence of the surveys, the biographical reading is not wrong about the man. It is wrong about what the man’s popularity proves.

A personal vote is not transferable capital for a party; it is closer to a tenancy. It holds the coalition together for the duration of the leader’s occupancy and it fills halls, and it does not appear in the party’s vote share, because electors who vote for an allied candidate out of regard for Vajpayee add nothing to the BJP’s column and owe the BJP nothing when the tenancy ends. The 1999 figures carry the signature of the tenancy with unusual clarity. The prime minister’s personal standing rose to its career height across the war summer, the campaign was designed around his person to a degree unusual even in leader-centred Indian elections, and the party’s vote share fell by 1.84 percentage points while its seat count stood still. If personal popularity were converting into partisan growth, the conversion would be visible somewhere in that pair of numbers. It is visible nowhere. The BJP of 1999 was a party borrowing its leader’s credit to hold an architecture together, and a balance sheet that records the loan as income will mislead its keeper in exactly the way the 2004 accounts were about to demonstrate.

The Kargil Attribution Problem

One decoration on the standard reading remains to be examined: the claim that the Kargil war won the election. The chronology is exact. The Vajpayee government lost its confidence vote on 17 April 1999 and continued in a caretaker capacity after President K. R. Narayanan dissolved the Lok Sabha on 26 April 1999. Pakistani intrusions across the Line of Control in the Kargil sector were detected in early May 1999, and the Indian armed forces fought Operation Vijay from late May until the intrusions were cleared by 26 July 1999. Polling followed from 5 September to 3 October 1999. The caretaker prime minister who had conducted the war, refusing to cross the Line of Control and winning international acceptance of India’s restraint, including President Clinton’s pressure on Prime Minister Nawaz Sharif at their Washington meeting of 4 July 1999, went before the electorate within weeks of the victory. The inference that the war delivered the verdict writes itself, and it was written, in the campaign and in much of the commentary that followed.

Why was a general election held in 1999 at all?

Because the second Vajpayee government lost its majority in the Lok Sabha. The AIADMK’s general secretary, J. Jayalalithaa, withdrew her party’s backing on 14 April 1999, and the confidence vote of 17 April 1999 went against the government by 269 votes to 270.

Jayalalithaa’s withdrawal followed the government’s refusal of demands that included the dismissal of the DMK government in Tamil Nadu, and the single-vote margin was supplied by the Bahujan Samaj Party’s late decision to vote with the opposition in the division. No alternative government could be assembled, and the House was dissolved.

The post-poll evidence disciplines the inference. The National Election Study of 1999, the post-poll survey conducted by Lokniti at the Centre for the Study of Developing Societies, asked voters what had shaped their choices, and the analyses that Yogendra Yadav and Suhas Palshikar built on that data, published across 1999 and 2000, found the war’s electoral imprint to be uneven in a way the national story does not suggest. Approval of the government’s handling of Kargil was broad, and Vajpayee’s personal standing was high in every region. Yet the BJP’s vote share fell nationally, and its seat tally in the northern states where the war’s salience was greatest did not rise: the party’s Uttar Pradesh contingent halved, from 57 seats to 29. A war effect that cannot be found in the winner’s own vote in the war’s most attentive region is an effect that commentary has overstated. Where the data do show movement consistent with a Kargil premium, it is concentrated in a band of Hindi-belt constituencies and frequently inseparable from alliance changes in the same seats, the Telugu Desam arrangement in Andhra Pradesh and the Janata Dal (United) consolidation in Bihar being the largest confounds.

The parliamentary record supports the narrower reading. In the Lok Sabha debates of October and November 1999, the opposition pressed the government on the intelligence failures that had preceded the intrusion, and the government resisted demands for a joint parliamentary committee, pointing instead to the Kargil Review Committee chaired by K. Subrahmanyam, appointed on 29 July 1999, whose report was laid before Parliament in February 2000. A government confident that the war had been its electoral trump might have been expected to welcome a parliamentary inquiry as a stage for vindication. The resistance suggests the victors themselves understood the war’s political value to be perishable and its scrutiny dangerous. The adjudication this article reaches is the measured one the CSDS evidence supports: Kargil raised Vajpayee’s personal stature and the caretaker government’s reputation for steadiness, both of which eased the alliance’s reassembly; the seats were supplied by the partners, and the space was supplied by the Congress’s collapse. Decoration is not foundation.

The war’s prehistory deepens the attribution problem rather than resolving it in the victors’ favour. In February 1999, two months before his government fell, Vajpayee had travelled by bus to Lahore and signed the Lahore Declaration with Prime Minister Nawaz Sharif, in the most publicised peace initiative of the decade between the two nuclear-armed neighbours. The Kargil intrusions, planned in the same months by the Pakistani military establishment, were therefore discovered by a government whose diplomatic signature achievement the intrusion betrayed, and the caretaker administration that fought the war from May to July 1999 was conducting military operations under a constitutional novelty commentators noted at the time: a government that had lost the confidence of the Lok Sabha, holding office only until elections could be held, exercising the war powers of the state. Vajpayee’s management of that position, prosecuting the war to the Line of Control and no further, declining to open additional fronts, and sustaining a diplomatic campaign that isolated Pakistan at Washington and in the G8’s statements, drew acknowledgement across the Indian political spectrum, including from the Congress benches in the parliamentary debates that followed. Credit of that breadth, however, attaches to the statesman and the state, and the survey evidence’s failure to find a partisan dividend is the more telling for the credit’s genuineness. The electorate distinguished, with a precision commentary has not always matched, between gratitude to a caretaker prime minister and allegiance to his party.

The Kargil Review Committee’s report, laid before Parliament in February 2000, performed a parallel distinction in the institutional register. The committee under K. Subrahmanyam documented the intelligence and surveillance failures that had left the intrusions undetected, and the debates of that winter in the Lok Sabha, in which the opposition pressed its demand for a joint parliamentary committee and the government defended the review committee’s sufficiency, turned on responsibility for the failures rather than on laurels for the recovery. A war whose official aftermath centred on the state’s unpreparedness is an unlikely engine for the victor’s party growth, and the committee’s findings stand in the record as the period’s own correction to the campaign-season inference. The Kargil effect, properly sized, was a personal premium for Vajpayee and a reputational asset for the alliance’s anchor. Sizing it as the cause of 303 seats requires the seats to have been won by the party the premium attached to, and the Election Commission’s tables show they were won, in the decisive increments, by other parties entirely.

The Constitutional Temptation: What the Government Thought Its Majority Licensed

The misreading had an institutional expression in the government’s first year, and it is worth recording because it shows the borrowed majority being spent, in imagination, on the largest questions a republic contains. In his address to the nation on the eve of Republic Day in January 2000, President K. R. Narayanan used the fiftieth anniversary of the Constitution to warn against haste in revising the founding settlement, remarks delivered as the government prepared its announced review of the Constitution’s working. The National Commission to Review the Working of the Constitution was appointed on 22 February 2000 under the chairmanship of M. N. Venkatachaliah, the former Chief Justice of India, with terms of reference the government framed as stocktaking after fifty years. The opposition, and a substantial body of constitutional commentary, received the announcement as something more ambitious: an attempt to reopen first principles, parliamentary government against presidential alternatives among them, on the authority of a majority its holders believed to be foundational. The commission’s eventual report of March 2002 recommended no such transformation, and the episode closed quietly.

The episode’s relevance here is diagnostic. Governments propose constitutional reviews for many reasons, and stocktaking at a golden jubilee is a respectable one. What the winter of 1999 to 2000 revealed was the government’s own estimate of its mandate’s depth, an estimate at odds with the Election Commission’s tables in the precise way this article has documented. A leadership that understood its majority as an assembly of rented state verdicts, held by a stagnant core party, would have husbanded its authority for the legislation its partners could sustain. A leadership that believed the 303 recorded a national conversion reached, within four months of taking office, for the Constitution itself. President Narayanan’s public caution, delivered from the republic’s highest office in the government’s first season, stands in the record as the earliest institutional correction offered to the misreading, and, like the state election results that followed it, it was received with courtesy and filed without effect.

The Lesson Mislearned: Confidence Without Growth, 1999 to 2002

A verdict misread becomes a strategy misbuilt, and the interval from 1999 to 2002 shows the misbuilding in progress. The government’s managers proceeded as though the 303 seats of 1999 represented a floor the party owned rather than a ceiling its partners had lent. No organisational drive comparable to the victory’s scale was undertaken in the states where the BJP’s own vote had fallen; the party’s Uttar Pradesh unit, steward of the halved contingent of 29, spent the years after 1999 in leadership churn, cycling chief ministers from Kalyan Singh to Ram Prakash Gupta to Rajnath Singh between November 1999 and 2002, and the state’s assembly election of February 2002 returned the BJP a distant third behind the Samajwadi Party and the Bahujan Samaj Party. A party reading 1999 correctly, as borrowed breadth over a stagnant core, would have treated those years as a race to build the core before the loan was recalled. The record shows the opposite disposition: the loan was treated as an asset, and the building was postponed.

The Gujarat violence of 2002 was managed inside the same misreading. After the burning of a train coach at Godhra on 27 February 2002 killed 59 passengers, most of them Hindu pilgrims and activists returning from Ayodhya, the retaliatory violence across Gujarat killed, on the Union government’s figure given to Parliament in May 2005, 1,044 people, the large majority of them Muslims, and displaced well over one hundred thousand. The state government of Narendra Modi was accused by the opposition, by civil liberties organisations and by the National Human Rights Commission, whose proceedings of March and April 2002 recorded its findings on the state’s failures, of complicity through inaction; the state government and the BJP rejected the charge and attributed the violence to spontaneous public anger, a characterisation their critics contested then and contest in the literature since. The facts of the death toll and the dispute over responsibility are both part of the record, and each is stated here in the terms its holders use. What belongs to this article’s argument is the political management at the centre. Vajpayee’s speech to the BJP National Executive at Goa on 12 April 2002, and his subsequent decision, with L. K. Advani, against replacing the Gujarat chief minister, were calculations made inside a confidence that the party’s national position could absorb the cost. A leadership that believed its majority was borrowed, and its core stagnant, would have priced the reputational cost differently. The BJP’s victory in the Gujarat assembly election of December 2002, with 127 of 182 seats, then fed the confidence further, converting a state verdict delivered in a polarised aftermath into apparent confirmation of national expansion.

The compounding error was analytical before it was moral or tactical. Each success of the architecture, the full term approaching, the Gujarat result, the economic indicators of 2003, was booked to the party’s account, because the founding misreading of 1999 had defined the alliance’s totals as the party’s strength. Organisations correct their readings when feedback contradicts them; the feedback of 2000 to 2002, the Uttar Pradesh decline, the stagnant by-election record, the dependence on partners for every parliamentary comfort, was available and was discounted. By the time the strategy for the next general election was set, the misreading was no longer a proposition anyone in the party’s leadership felt obliged to defend. It had become the premise.

The Feedback Nobody Read: State Verdicts from 2000 to 2003

Between the misreading and its price stood four years of corrective evidence, delivered in the ordinary currency of state elections, and the record of those years shows the evidence being received and filed without effect. The assembly elections of 2001 returned Congress-led or Congress-supported outcomes in Assam and Kerala and a Left Front renewal in West Bengal, while the BJP’s own state presence outside its strongholds remained dependent on partners. The Punjab assembly election of February 2002 installed a Congress government under Amarinder Singh at the expense of the BJP’s Akali partners. Himachal Pradesh followed with a Congress victory in 2003. None of these results was a national verdict, and each could be, and was, explained away in New Delhi as local. Their cumulative testimony, that the Congress’s organisational recovery under Sonia Gandhi was proceeding state by state while the BJP’s core electorate showed no corresponding enlargement, was the testimony the 1999 figures had already given. It was discounted because the premise admitted no contrary evidence.

The Tehelka exposure of March 2001, in which journalists posing as arms dealers recorded defence-procurement conversations that reached the BJP’s president, Bangaru Laxman, and forced his resignation, together with the temporary departure of George Fernandes from the defence ministry, supplied a second species of feedback: the government’s reputation for probity, a component of the Vajpayee premium on which the whole architecture rested, was neither self-sustaining nor within the party’s control. The Unit Trust of India’s US-64 crisis of 2001 added a third, in the domain of middle-class economic confidence that the government’s managers counted as their strongest ground. Each episode was managed, as episodes are. None was read as data about the fragility of a majority whose foundations were a leader’s reputation and a set of rented state organisations, because the founding misreading had already classified the majority as the party’s property. The feedback nobody read is a standard feature of verdicts misread at the moment of triumph; the 1999 case is instructive because the misreading is so cleanly recoverable from the Election Commission’s own tables, which sat in print the entire time.

The Repair Shop: How the Congress Rebuilt Between 1999 and 2004

The final piece of evidence for the adjudication is the behaviour of the loser, because the Congress after 1999 acted on a diagnosis of its defeat that matched this article’s reading point for point. The party appointed a committee under A. K. Antony to inquire into the 1999 debacle, and the inquiry’s circulation inside the party through 2000 identified the causes the seat arithmetic suggests: organisational decay in the states, the absence of alliances, the failure to project state-level leadership. Parties that believe they lost to an opponent’s charisma do not commission organisational autopsies. The Congress’s own post-mortem treated 1999 as a machinery failure, and the five years that followed were spent, with setbacks and resistance, on the machinery.

The resistance is part of the record. In November 2000, Jitendra Prasada, a vice-president of the party from Uttar Pradesh, challenged Sonia Gandhi for the Congress presidency in the first contested election for the office in decades, and was defeated by a margin that demonstrated the organisation’s preference for the incumbent’s consolidating hand over a return to the factional market. The contest matters here because it was the old Congress’s last institutional stand: the provincial bosses’ candidate, running on the grievances of the displaced networks, could not assemble a coalition inside his own party, in an echo of the national failure of April 1999. Thereafter the rebuilding proceeded on the president’s terms: organisational appointments recentralised, state leaderships in the major states gradually replaced or confirmed, and, most consequentially, the party’s doctrine on alliances reversed. The Congress session at Shimla in July 2003 adopted the position, formalised in its Shimla Sankalp, that the party would enter pre-poll arrangements with like-minded formations against the BJP, an abandonment of the solitary posture that had produced both the April 1999 humiliation and the 114 seats. Through the winter of 2003 to 2004 the arrangements followed: reconciliation with the Nationalist Congress Party in Maharashtra, an understanding with the Rashtriya Janata Dal in Bihar, the Dravida Munnetra Kazhagam’s transfer of allegiance in Tamil Nadu, and state-level seat divisions across the map. The alliance that won in May 2004 was the photographic negative of the campaign that had lost in October 1999, assembled by the same leader whose inexperience the earlier defeat had been attributed to. The variable that changed between the two elections, on the Congress side, was not the leader’s person. It was the organisation and the diplomacy, the two capacities whose absence the collapse reading identifies as the substance of 1999.

Feel Good Resolutions and the India Shining Campaign

The premise met its test in the winter of 2003 to 2004. The BJP’s assembly victories in Madhya Pradesh, Rajasthan and Chhattisgarh, counted on 4 December 2003, were real achievements of state organisation and anti-incumbency, delivered under Uma Bharti, Vasundhara Raje and Raman Singh. Read through the 1999 lens, they were bound to be misjudged in one direction: as evidence of a national mood for the BJP rather than as state verdicts of the ordinary alternating kind. The party’s national deliberations at the turn of 2004 adopted the optimistic reading. The Union cabinet recommended dissolution of the Lok Sabha, the House was dissolved on 6 February 2004, and the general election was advanced by roughly half a year, to April and May 2004, on the explicit logic that conditions favoured the incumbent alliance as they had not favoured it since the victory of 1999.

The campaign built on that logic was India Shining, a publicity programme originated by the government’s own advertising in 2003 and adopted as the BJP’s electoral theme, celebrating high growth figures, foreign exchange reserves and infrastructure construction as the lived experience of the ordinary elector. Its content is the final proof of the misreading this article has traced. A campaign celebrating aggregate prosperity presumes an electorate that credits the ruling party with that prosperity and intends to reward it, which is to say it presumes exactly the breadth of partisan attachment that the 1999 figures had denied. The Congress-collapse reading would have counseled a different campaign: defensive, constituency-specific, priced on the assumption that allied votes in Andhra Pradesh, Bihar and Tamil Nadu were held by partners whose own fortunes were shifting, and that the BJP’s core required mobilisation rather than congratulation. Chandrababu Naidu’s Telugu Desam Party faced a resurgent Congress in Andhra Pradesh; the partners’ state-level troubles were visible in advance and were discounted in New Delhi in the same currency as every earlier warning.

The opposition, by contrast, campaigned on the correct reading of 1999 without needing to state it. Sonia Gandhi’s Congress spent 2003 and early 2004 doing the organisational work whose absence had destroyed it in 1999: negotiating state-level arrangements, accommodating regional formations, and presenting an alliance, the United Progressive Alliance in embryo, whose architecture mirrored the method that had beaten it. The party whose collapse had supplied the 1999 majority had repaired the specific deficiency the collapse exposed. The party that had misread the collapse had repaired nothing, because it did not believe anything was broken.

The Readers’ Quarrel: Yadav, Palshikar, Jaffrelot and Malhotra on 1999

The disagreement this article adjudicates was conducted in print by named readers in the verdict’s immediate aftermath, and their positions deserve individual statement, because the article’s claim is a judgement among them rather than a summary of all. Yogendra Yadav, writing from the National Election Study data in his analyses of 1999 and in the framework he called the post-Congress polity, read the verdict as a consolidation: the party system that had replaced the Congress System after 1989 had settled into a competitive pattern in which state-level contests, aggregated through alliances, determined national outcomes, and in which the BJP and the Congress stood as two poles neither of which could command the system alone. Suhas Palshikar’s state-level studies, developed across the same years from the Maharashtra evidence and generalised in his later formulations of the post-Congress polity, pressed the complementary point that the operative units of Indian electoral politics had become the states, each with its own party system, so that a national verdict was an arithmetic of twenty-odd distinct contests rather than a single national conversation. On the Yadav and Palshikar readings, both grounded in the survey and constituency evidence, the question the standard reading asks, whether the nation moved toward the BJP in 1999, is malformed: nations do not vote in India, state electorates do, and the state electorates of 1999 moved in contrary directions that the alliance totals then flattened into a single misleading number.

Christophe Jaffrelot’s reading, in the essays gathered in The BJP and the Compulsions of Politics in India in 2000 and in his subsequent studies of the party, supplied the BJP-side corollary. The party’s rise across the 1990s was, for Jaffrelot, a genuine sociological expansion bounded by a ceiling: its upper-caste core and its Hindutva programme limited its independent reach, and its national majorities were necessarily coalition artefacts, assembled by conceding to partners the programme’s deferral and the states’ autonomy. Inder Malhotra’s biographical writing on Vajpayee represents the fourth position, the leader-centred account in which the era’s hinge is a person: Vajpayee’s stature, moderation and parliamentary authority made him acceptable to partners and voters his party could not win, and the 1999 verdict, on this account, was the electorate’s ratification of the man. The adjudication entered in this article accepts the evidence of the first three readings and the biographical reading’s central fact, Vajpayee’s personal premium, which the survey data confirm. It rejects the inference the biographical reading invites, from the man’s premium to the party’s strength, because the premium is precisely what the Election Commission’s party-level figures fail to register anywhere except in the alliance’s aggregate. Where the four readers are weighed against the primary record, the strength interpretation of 1999 survives only in the form in which it was never actually advanced by any of them: as the victors’ internal folklore, the resolutions’ self-congratulation hardened into strategy. The scholarly quarrel and the party’s quarrel with its own future were the same quarrel, and the scholars won it five years before the party conceded.

A Note on Method: How a Composite Verdict Should Be Read

The argument of this article rests on a method, and the method should be stated plainly so that any reader can apply it to the Election Commission’s published tables and check the result. A general election verdict in India is a composite of state verdicts, and a composite must be disaggregated before it is interpreted. The disaggregation applied here has four steps. The largest party’s seats are compared with its own seats at the previous election, never with its opponent’s seats alone. Its vote share is compared with its own previous share, since a party’s national percentage is the closest available measure of its unaided appeal. The allied component of any alliance total is separated from the component the leading party won with its own candidates, and the allied component is then traced party by party and state by state, because allied seats answer to allied electorates. The losing party’s result is examined for the signature of conversion failure, a rising or stable vote producing a falling seat count, which distinguishes organisational collapse from electoral repudiation. Applied to the statistical reports of 1998 and 1999, the four steps return the same answer at every stage, which is the ground on which this article’s confidence rests.

Two cautions belong to the method, and both have been observed in the argument. The first concerns survey evidence. Post-poll surveys such as the National Election Study depend on respondents’ recollections of choices already made, and their findings on motivation, the weight of a war, the pull of a leader, carry wider margins of uncertainty than their findings on reported vote. This article has used the survey evidence for the questions it answers best, the distribution of preferences and the geography of movement, and has declined to rest any load-bearing claim on a motivational estimate. The second caution concerns the counterfactual. To call 1999 a Congress collapse is not to claim the BJP could not have won a well-contested election, nor that a repaired Congress would certainly have prevailed; the Congress’s vote share in 1999, even converted at the rates its healthier state units achieved, would have left the alliance competitive in a House where the Telugu Desam and the Janata Dal (United) held the balance. The claim is about what the verdict demonstrated, not about what might have happened in a different campaign. Verdicts are evidence about the parties that contested them, and the discipline of reading them as evidence, rather than as omens or as laurels, is the whole of the method. It is also the discipline the BJP’s strategists of 2003 and 2004 most conspicuously lacked, since a party reading its own composite honestly would have counted its owned seats, priced its rented ones, and campaigned for the election it actually faced.

The 2004 Verdict as Audit: The Adjudication

The verdict counted in May 2004 audited every claim made for 1999, and its line items can be read off the Election Commission’s statistical report on the 2004 general election with the same discipline applied throughout this article. The BJP fell from 182 seats to 138. The Congress rose from 114 to 145 and formed the government, with Manmohan Singh sworn in as prime minister on 22 May 2004 at the head of the United Progressive Alliance, sustained by the Left parties’ backing from outside. The Telugu Desam Party, the largest single contributor to the 1999 ledger, was swept from power in Andhra Pradesh in the simultaneous assembly election and reduced to 5 Lok Sabha seats, its state electorate having delivered the verdict on Chandrababu Naidu’s government that the alliance’s national managers had assumed away. The partners whose seats had been booked as the BJP’s growth in 1999 departed on schedules set by their own state politics, exactly as the Congress-collapse reading of 1999 predicted they would, because they had never been the BJP’s to keep.

Adjudication can therefore be entered between the two readings with which this article began. The BJP-strength reading held that 1999 registered the party’s durable national expansion, ratified by stability in office. Its falsifiable implications were that the party’s vote and seats would hold or grow at the next test, that allied support rested on partisan convergence, and that a campaign pitched on the government’s record would find the broad electorate the reading presumed. Each implication failed in 2004. The Congress-collapse reading held that 1999 registered the Congress’s organisational disintegration, the BJP’s seat-count stability, and an alliance expansion rented from regional formations. Its implications were that the BJP remained exposed wherever its own organisation was weak, that partners would follow state-level incentives, and that the Congress, if it repaired its machinery and its alliance diplomacy, would recover its position as the default alternative pole. Each implication was borne out: the Congress’s recovery was built in precisely the two currencies, state organisation and pre-poll arrangement, whose absence had produced 114 seats in 1999.

The claim this article defends can be stated in its final form. The 1999 Vajpayee victory was substantively a Congress collapse; the BJP’s seat-count stability is the giveaway. The sentence is not a diminishment of the victors’ genuine accomplishments, the war conducted under restraint, the government carried through its term, the economic legislation of 1999 to 2003. It is a correction of the ledger in which those accomplishments were recorded. A party that mistakes its opponent’s collapse and its partners’ rentals for its own growth will set its strategy by the mistaken figure, and the BJP’s managers did so from the National Council resolutions of 1999 to the dissolution of February 2004. The 2004 verdict was not a reversal of the meaning of 1999. It was the meaning of 1999, arriving five years late, in the only form the party that misread it could no longer discount.

There is a wider lesson for readers of Indian elections, and it is the lesson the coalition era teaches in election after election. Headline totals in a polity of state-level party systems are composites, and a composite can rise while every component the headline-writer cares about stands still or falls. The discipline the 1999 verdict rewards is disaggregation: the largest party’s seats against its own previous count, its vote share against its own previous share, the allied component separated from the owned component, and the loser’s machinery examined before the winner’s appeal is celebrated. Applied to 1999, the discipline yields the reading defended here. Applied with the same severity to the victors of later elections, it remains the surest protection against drawing, from any single verdict, the wrong lessons at the price the BJP paid for its lessons of 1999.

Frequently Asked Questions

Q: What was the 1999 Vajpayee victory?

The 1999 Vajpayee victory was the result of the general election held from 5 September to 3 October 1999, in which the Bharatiya Janata Party and its pre-poll allies in the National Democratic Alliance won 303 Lok Sabha seats, and Atal Bihari Vajpayee formed a government on 13 October 1999. The Election Commission of India’s statistical report records the BJP itself winning 182 seats, exactly its 1998 tally, on a reduced vote share of 23.75 percent; the Congress fell to 114 seats, its lowest count since independence. The victory is therefore best understood as an alliance expansion built over a Congress collapse, rather than as growth in the BJP’s own electoral strength.

Q: How many seats did the BJP win in 1999?

The Bharatiya Janata Party won 182 Lok Sabha seats in the 1999 general election, according to the Election Commission of India’s statistical report. The figure is the analytical centre of the verdict, because it is identical to the 182 seats the party had won in the general election of 1998, even though the BJP fought the 1999 election behind a prime minister at the height of his popularity, in the aftermath of the Kargil war, and at the head of a much broader alliance. The party also contested fewer seats, 339 against 388 in 1998, standing aside for its partners. Every additional seat in the alliance’s total of 303 was won by an allied party, not by the BJP.

Q: Did the Kargil war help the BJP in 1999?

The Kargil war helped Atal Bihari Vajpayee personally more than it helped his party measurably. The war, fought from May to 26 July 1999 against Pakistani intrusions in the Kargil sector during the months Vajpayee headed a caretaker government, raised his standing as a steady national leader, and the National Election Study post-poll analyses by the Centre for the Study of Developing Societies recorded broad approval of his government’s conduct of the conflict. Yet the BJP’s vote share fell from 25.59 percent in 1998 to 23.75 percent in 1999, and its seat count in Uttar Pradesh, where the war’s salience was high, dropped from 57 to 29. Analysts working from the survey data, including Yogendra Yadav, concluded that the war’s partisan dividend was concentrated in limited regions and confounded with alliance changes, rather than spread across the national verdict.

Q: Why did the Congress lose in 1999?

The Congress lost in 1999 principally because its organisation had decayed past the point at which its vote could be converted into seats. The Election Commission recorded the party polling 28.30 percent of the national vote, the largest share won by any single party, yet taking only 114 seats, its lowest tally since independence. The causes were structural: state units hollowed out during the Rao and Kesri years, organisational elections displaced by nomination, the absence of pre-poll alliances while the National Democratic Alliance pooled the anti-Congress vote behind single candidates, and a leadership succession, Sonia Gandhi’s accession in March 1998, too recent to have rebuilt the machinery. The party’s own post-defeat inquiry under A. K. Antony reached a compatible organisational diagnosis.

Q: Was Sonia Gandhi Congress President in 1999?

Yes. Sonia Gandhi became president of the Indian National Congress on 14 March 1998, when the Congress Working Committee removed Sitaram Kesri and installed her in his place, and she led the party through the 1999 general election. Her presidency was eighteen months old when the election was called, and it had already survived two defining tests: the failure in April 1999 to assemble an alternative government after the Vajpayee ministry fell by one vote, and the May 1999 expulsion of Sharad Pawar, P. A. Sangma and Tariq Anwar, who had challenged her eligibility to lead on grounds of her Italian birth and who founded the Nationalist Congress Party. She contested the 1999 election from Amethi and Bellary and won both seats.

Q: How long did the Vajpayee government last?

The government formed after the 1999 election, Atal Bihari Vajpayee’s third ministry, lasted from 13 October 1999 until 22 May 2004, a span of four years and seven months. It is frequently described as the first government led by a non-Congress prime minister to complete a full term, a description that holds in substance, since the ministry ran the effective course of the thirteenth Lok Sabha; the House was dissolved on 6 February 2004, some months ahead of the full five years, when the government chose to advance the general election to April and May 2004 in the confidence generated by the India Shining campaign. The decision to end the term early is part of the verdict’s story, since it rested on the misreading of 1999 examined in this article.

Q: How did 1999 lead to the 2004 election?

The 1999 verdict led to the 2004 outcome through the lesson the victors drew from it. Reading the alliance’s 303 seats as evidence of the BJP’s own expansion, although the party’s seats had not moved and its vote share had fallen, the government governed and planned as though its majority were owned rather than assembled from partners. The 2002 Gujarat violence was managed without pricing its national reputational cost, the feedback of state elections from 2000 to 2003 was discounted, and the assembly victories of December 2003 were treated as proof of a national mood. The Lok Sabha was dissolved early, on 6 February 2004, and the India Shining campaign presumed a breadth of partisan support the 1999 figures had never shown. In May 2004 the BJP fell to 138 seats and the Congress, rebuilt organisationally and diplomatically, formed the government.

Q: How many seats did the National Democratic Alliance win in 1999?

The National Democratic Alliance won 303 Lok Sabha seats in the 1999 general election, a comfortable majority in the 543-member House. The composition of the total matters more than the total itself. The Bharatiya Janata Party contributed 182 of the 303, and allied parties contributed 121, up from the 70 seats allied parties had won alongside the BJP in 1998. The largest allied contingents in the Election Commission’s statistical report were the Telugu Desam Party with 29 seats, the Janata Dal (United) with 21, the Shiv Sena with 15, the Dravida Munnetra Kazhagam with 12 and the Biju Janata Dal with 10. The alliance’s growth between the two elections was entirely a growth in the allied component, which is the numerical foundation of the Congress-collapse reading of the verdict.

Q: How far did the Congress fall between 1998 and 1999?

The Congress fell from 141 Lok Sabha seats in the general election of 1998 to 114 seats in 1999, a loss of 27 seats in eighteen months, according to the Election Commission’s statistical reports on the two elections. The fall carried the party below its previous historical floor, the 154 seats of the 1977 verdict that followed the Emergency. The most instructive feature of the fall is its relationship to the party’s vote: the Congress’s national vote share rose over the same interval, from 25.82 percent to 28.30 percent, the largest share recorded by any single party in 1999. A party gaining votes and losing a fifth of its seats is suffering a failure of conversion, produced by disorganisation and alliancelessness, rather than a rejection by the electorate.

Q: What vote share did the BJP record in 1999 compared with 1998?

The Bharatiya Janata Party recorded a vote share of 23.75 percent in the 1999 general election, down from 25.59 percent in 1998, on the Election Commission of India’s statistical reports for the two elections. The decline of 1.84 percentage points accompanied an unchanged seat count of 182, an unusual combination explained by the party’s narrower candidature: the BJP contested 339 seats in 1999 against 388 in 1998, withdrawing in favour of allied candidates across large parts of the map. Its strike rate consequently improved while its national footprint contracted. The vote share figure is the single most direct statistical contradiction of the claim that the 1999 verdict registered a growth in the BJP’s own electoral strength.

Q: Why did Sharad Pawar, P. A. Sangma and Tariq Anwar break with the Congress in 1999?

The three leaders broke with the Congress over the question of Sonia Gandhi’s eligibility to lead the party and, by implication, to become prime minister. On 15 May 1999, weeks after the fall of the Vajpayee government, Sharad Pawar, P. A. Sangma and Tariq Anwar wrote to the Congress president arguing that the party should be led by a person born in India, a challenge framed around her birth in Italy. The Congress Working Committee treated the letter as indiscipline and expelled all three on 20 May 1999. The expelled leaders founded the Nationalist Congress Party on 25 May 1999, and the new party won 8 Lok Sabha seats that autumn, concentrated in Maharashtra, where Pawar’s network had been the Congress’s strongest surviving state organisation. The split materially deepened the parent party’s organisational crisis on the eve of the election.

Q: Which parties entered the BJP-led alliance for the first time in the 1999 election?

The most consequential first-time entrants to the BJP-led alliance in 1999 were the Dravida Munnetra Kazhagam in Tamil Nadu and the Telugu Desam Party in Andhra Pradesh, together with the newly assembled Janata Dal (United), formed that year from the Samata Party, Janata Dal fragments and the Lok Shakti. The Dravida Munnetra Kazhagam, which won 12 seats, had never previously allied with the BJP and entered the arrangement its state rival, the AIADMK, had vacated by withdrawing from the government in April 1999. The Telugu Desam Party, which won 29 seats, sustained the alliance from outside the cabinet. The Indian National Lok Dal in Haryana, with 5 seats, was another addition of the 1999 round. Each entrant joined on state-level calculations, and each retained the freedom, exercised by several within the term, to renegotiate or leave.

Q: How did the Telugu Desam Party sustain the Vajpayee government after 1999?

The Telugu Desam Party sustained the Vajpayee government from outside the council of ministers. Chandrababu Naidu, the party’s leader and the Chief Minister of Andhra Pradesh, declined cabinet berths for his party after the 1999 election and instead supported the National Democratic Alliance government externally, voting with it in the Lok Sabha and preserving his freedom to bargain issue by issue. The Election Commission recorded the Telugu Desam winning 29 seats, the largest allied contingent in the alliance, which made its external backing the government’s single largest prop. The arrangement was symbolised by the election of the Telugu Desam’s G. M. C. Balayogi as Speaker of the Lok Sabha. Naidu used the leverage the arrangement conferred to press Andhra Pradesh’s claims on the centre, and his party’s eventual defeat in the Andhra Pradesh assembly election of 2004 removed the prop at the moment of the government’s own defeat.

Q: What did the National Election Study of 1999 find about the way voters had moved?

The National Election Study of 1999, the post-poll survey conducted by Lokniti at the Centre for the Study of Developing Societies, found a verdict of consolidation rather than conversion. The analyses published by Yogendra Yadav, Suhas Palshikar and their colleagues across 1999 and 2000 showed state electorates moving in different directions according to state-level contests, with no uniform national swing toward the Bharatiya Janata Party, whose vote share the Election Commission recorded as falling. The survey registered Atal Bihari Vajpayee’s large personal lead in the electorate’s preference for prime minister, broad approval of the government’s conduct of the Kargil war, and constituency choices governed heavily by state governments’ records and alliance configurations. The data supported reading the verdict as an aggregation of state outcomes sorted by alliance arithmetic, the interpretation from which the Congress-collapse analysis of 1999 proceeds.

Q: How did the BJP perform in Uttar Pradesh in 1999 compared with 1998?

The Bharatiya Janata Party’s performance in Uttar Pradesh deteriorated sharply between the two elections. In 1998 the party won 57 of the state’s 85 Lok Sabha seats, the foundation of its national tally. In 1999 it won 29, a loss of 28 seats in its principal stronghold in the very election its national alliance total rose to 303. The Election Commission’s statistical reports record the Samajwadi Party winning 26 seats in the state in 1999 and the Bahujan Samaj Party 14, with the Congress reduced to 10. The Uttar Pradesh figures are among the strongest single pieces of evidence against the BJP-strength reading of the 1999 verdict, since a party expanding nationally does not ordinarily halve its contingent in the largest state while its leader’s popularity stands at its peak.

Q: Who was the National Democratic Alliance’s convenor in 1999, and what did the office do?

George Fernandes, the leader of the Samata Party and subsequently a principal of the Janata Dal (United), was the convenor of the National Democratic Alliance in 1999. The office was the coalition’s working mechanism: the convenor chaired the alliance’s coordination among its constituent parties, negotiated the distribution of portfolios and the terms of the common programme, and managed the recurring disputes through which a majority assembled from more than twenty formations had to be maintained. Fernandes held the Defence portfolio in the government formed on 13 October 1999, combining the alliance’s internal management with one of its most sensitive ministries. The prominence of the convenor’s office in the 1999 arrangements is itself evidence of the verdict’s character, since a majority owned by a single party has no need of a standing negotiator to hold it together.

Q: What happened to the AIADMK between the elections of 1998 and 1999?

The AIADMK travelled from the government’s indispensable partner to the opposition’s camp within a single electoral cycle. In the 1998 election the party won 18 Lok Sabha seats in alliance with the BJP, and its general secretary, J. Jayalalithaa, became the second Vajpayee government’s most demanding supporter, pressing demands that included the dismissal of the DMK government in Tamil Nadu. On 14 April 1999 she withdrew her party’s backing, and the government lost the confidence vote of 17 April 1999 by 269 votes to 270. In the 1999 general election the AIADMK fought in arrangement with the Congress and won 10 seats, while its state rival, the Dravida Munnetra Kazhagam, took the AIADMK’s former place inside the National Democratic Alliance and won 12. The exchange is the clearest illustration in the 1999 record of alliance breadth as brokerage rather than partisan growth.

Q: When was the 1999 general election held, and when did the new government take office?

The 1999 general election was held in five phases from 5 September to 3 October 1999, with counting on 6 October 1999, according to the Election Commission of India’s record of the poll. The election had been necessitated by the dissolution of the twelfth Lok Sabha on 26 April 1999, after the second Vajpayee government lost a confidence vote by a single vote on 17 April and no alternative government could be formed. Atal Bihari Vajpayee, who had headed the caretaker administration through the intervening Kargil war, took the oath as prime minister for the third time on 13 October 1999, a week after the counting, at the head of the National Democratic Alliance majority of 303 seats.

Q: What did Christophe Jaffrelot argue about the BJP after the 1999 verdict?

Christophe Jaffrelot argued, in the essays collected in The BJP and the Compulsions of Politics in India, published in 2000 with Thomas Blom Hansen, and in his related studies of the period, that the Bharatiya Janata Party’s expansion across the 1990s was genuine but bounded. On his reading, the party’s social base and its Hindutva programme imposed a ceiling on its independent electoral reach, and its parliamentary majorities were necessarily coalition artefacts, assembled by deferring the programme’s core commitments, on Ayodhya, Article 370 and a uniform civil code, and by accommodating regional partners whose votes the party did not command. The 1999 figures fit his analysis closely: the party’s seats were unchanged at 182, its vote share fell, and the majority materialised in the accounts of its allies. His reading supplies the scholarly form of this article’s Congress-collapse argument from the BJP’s side of the ledger.

Q: How did the state election results of December 2003 shape the decision to advance the 2004 election?

The Bharatiya Janata Party’s victories in the Madhya Pradesh, Rajasthan and Chhattisgarh assembly elections, counted on 4 December 2003, persuaded the government’s managers that a favourable national moment had arrived and should be seized before it faded. Read as evidence of expanding national support for the party, rather than as state verdicts driven by anti-incumbency against Congress state governments, the results encouraged the Union cabinet to recommend an early dissolution, and the Lok Sabha was dissolved on 6 February 2004, advancing the general election by roughly half a year to April and May 2004. The India Shining campaign was built for the advanced timetable. The decision is the terminal consequence of the 1999 misreading examined in this article: a party that had mistaken allied seats for its own growth in 1999 mistook three state victories for a national mandate in 2003, and scheduled its reckoning accordingly.