Abstract
One of the most important events in U.S. congressional history is the 1910 revolt against Speaker Cannon. The rebellion had myriad ramifications for the inner workings of the House of Representatives and dramatically altered the chamber’s power structure. Despite its significance, we do not have a clear understanding of the character of the revolt and why the 42 Republican insurgents revoked their allegiance to Cannon and their party. Did the insurgents rebel because of their strong progressive ideals or for more pragmatic reasons, such as political survival or retribution? Using data gathered from Cannon’s personal papers and other sources, we systematically explore disparate explanations credited for the revolt. For the progressive core of the insurgency, our analysis indicates that policy differences drove their behavior. These early insurgents were later joined by a group of less progressive members who appear to have supported the rebellion for electoral and retaliative reasons.
The 1910 revolt against Speaker Cannon was a watershed in congressional history. The rebellion initiated a series of decisions that would dramatically alter power in the House of Representatives. Despite its significance, we do not have a clear understanding of the character of the revolt and why the 42 Republican insurgents revoked their allegiance to Speaker Cannon and their party. Did the insurgents rebel because of their strong progressive ideals or for more pragmatic reasons, such as political survival or retribution for Cannon’s heavy-handed control of the chamber? Given the gaps in our knowledge, our insight into what prompted this rebellion is limited.
In this article, we systematically explore several explanations for the revolt against Cannon. Using data gathered from Cannon’s personal papers and other sources, we consider why the Republican insurgents behaved as they did. Specifically, we examine whether the rebellion was motivated by ideological principles, retribution for unjust committee assignments or undemocratic control of the legislative agenda, or simple electoral survival.
The Insurgents and the Insurgency
The revolt against Cannon climaxed with a favorable vote on the Norris resolution in March 1910. The resolution created an enlarged and elected Rules Committee on which the speaker would not be allowed to serve. It was a decision that was substantively of limited importance—Norris (1961, p. 121) admitted in his autobiography, “I think the country gave us credit for more than actually was accomplished in reform”—but in historical hindsight it represented a political earthquake. For the first time the Republican insurgents had managed to defeat the speaker on a matter both sides saw as significant. But it did not overthrow the speaker; a few days later the House declined Cannon’s offer to resign.
The rebellion had long been simmering. News stories about the “despotic” speaker being opposed by Republican “insurgents” began appearing in 1904 (New York Times, 1904). A year later a GOP representative from Wisconsin wrote to his brother that “There is a pretty strong revolt brewing against the domination of the House leaders and the efforts to crucify every man who does not bow to the dictates of . . . the Speaker of the House” (Harrison, 1992, p. 3). By 1908 a number of Republican House nominations were being contested between Cannon supporters and opponents (Griffith, 1965; National Tribune, 1908; New York Times, 1908c; Washington Times, 1908c). Many candidates, among them E. A. Hayes (Republican, CA), Irvine Lenroot (Republican, WI), and George Norris (Republican, NE), ran explicitly anti-Cannon campaigns (Margulies, 1977, pp. 72-76; McCook Tribune, 1908b; New York Times, 1908b; San Francisco Call, 1908; Washington Herald, 1908). Such behavior did not go unnoticed: In his personal papers, Cannon made a point to record that Miles Poindexter (Republican, WA) had campaigned as an anti-Cannon candidate. More important, these were not isolated events. One newspaper reported, “The Cannon blight is playing havoc with a large part of the crop of Republican candidates for re-election. . . . Although symptoms of the devastating taint are found in all parts of the country, they are especially marked in the middle Western states . . . (Washington Times, 1908a).
As their grievances accumulated, in December 1908 the insurgents opted to formally organize. The representatives who met were motivated by a desire to break the speaker’s power. But each attendee had a somewhat different notion of what needed to be done and how it might be accomplished. The one thing on which they agreed was that the power to appoint committees had to be taken from the speaker (Atkinson, 1911, p. 95; New York Tribune, 1908a, 1908b; Norris, 1910, p. 7).
In early 1909, 29 Republican House members submitted a resolution to amend the chamber’s rules. The proposal’s major thrust was to remove the speaker from the Rules Committee and to assign the power to make committee assignments to a new Committee on Rules and Committees. The 29 insurgents behind the proposal were not a random sample of House Republicans. Instead, they were disproportionately drawn from the upper Midwest, with five members from Wisconsin, five from Nebraska, four from Minnesota, three from Iowa, and two from North Dakota. In La Follette’s Weekly Magazine, progressives fêted this group with “a role of honor.” Another seven House members were acknowledged as “near-insurgents” for their public support of more limited rules reforms (“Reform of the House Rules,” 1909). The resolution only failed when over 20 Democrats bolted their party to support Speaker Cannon (“Party Machines Uphold Cannonism,” 1909).
The speaker delayed making most committee assignments until August (Hasbrouck, 1927, p. 37), and at that point, he took his revenge on his recalcitrant colleagues. As the New York Times reported, “Speaker Cannon drove the steam roller today over such of the insurgent Republicans as had acquired prominent committee assignments.” In the reshuffling, Henry Cooper (Republican, WI) lost the chair of Insular Affairs, Charles Fowler (Republican, NJ) was replaced as chair of Banking and Currency, and Augustus Gardner (Republican, MA) was removed as chair of Industrial Arts and Expositions. Gardner was actually an interesting case. The speaker liked him personally and wanted to spare him, but after hearing about the impending punishments the Massachusetts representative asked Cannon that he be treated like the other insurgents. Many of those who did not hold chairs also paid a price. For instance, William Lovering (Republican, MA), a seven-term veteran, was removed from Interstate Commerce and exiled to Manufactures, “which never holds a meeting,” and Coinage, Weights, and Measures. And both Lenroot and Poindexter were banished to Ventilation and Acoustics. But not every insurgent suffered and a few even managed to come out ahead: Edmond Madison (Republican, KS) and Moses Kinkaid (Republican, NE) retained their assignments, and James Good (R-Iowa) got moved to several “desirable” committees (“Cannon’s Revenge,” 1909; New York Times, 1909; Washington Times, 1909b).
Skirmishes between the speaker and the insurgents continued into the early months of 1910. Occasionally, the rebels won a battle. In January, they joined with Democrats to take away the speaker’s power to appoint six House members to a joint committee to investigate the Ballinger–Pinchot affair. In a more minor rebuke, the House voted to deny the speaker a touring car, forcing him to continue to use a horse-drawn carriage. One insurgent joked that “Our ‘Uncle Joe’ is dangerous enough when he is sitting still, presiding over the House . . . without our taking any chances of his overturning us and running us to cover outside” (New York Times, 1910c). Washington was consumed by the ongoing conflict. It even surfaced at the Gridiron Club dinner, which featured a skit where a pack of insurgents escaped from “Uncle Joe’s Cabin” (New York Times, 1910b). Thus, no one was surprised when shortly thereafter Norris seized an opportunity to challenge the speaker’s powers.
This brief synopsis of a long and intricate series of events suggests three reasons why the insurgents might have decided to challenge a speaker from their own party. First, they thought that the speaker’s powers prevented them from being able to pursue their policy preferences. This unhappiness was manifested on two related dimensions. One was procedural. The insurgents thought that the speaker’s powers were undemocratic. Along these lines, a Wisconsin Republican House member, John Nelson, publicly called Cannon “the American Frankenstein” because of “the arbitrary power of the Speaker” (Nelson, 1909). A Kansas colleague, Victor Murdock, concurred, claiming, “The House has surrendered its power to the presiding officer,” and become a place “where no man can get his measure up unless he sees the Speaker first; where he can’t gain the floor without being ‘in’ with the machine” (Murdock, 1909; Washington Times, 1908d). A Republican representative from Iowa, Elbert Hubbard, lamented that under Cannon, the House had degenerated into “a little oligarchy under the control of a few men” (Cedar Rapids Evening Gazette, 1908b).
The other dimension was policy. The GOP dissidents favored different policies than those backed by the speaker. For instance, Murdock first got involved in the anti-Cannon movement because the speaker thwarted his efforts to reform certain untenable postal service practices, changes President Theodore Roosevelt ultimately had to carry through by executive order (La Forte, 1974, pp. 138-139). Charles Lindbergh of Minnesota, perhaps the most radical insurgent (and father of the famed aviator), fought Cannon because he thought the speaker impeded progressive legislation and organized the House “to the advantage of a certain few” (Larson, 1973, p. 73). Complaints over procedure and policy melded into the insurgents’ overarching objection: “The system under attack has the evil portion of its life in the successful denial of vital participation on the part of the majority in constructive legislation” (Murdock, 1910, p. 511). The idea that the insurgents were motivated by their disenchantment with Cannon’s procedural control over the policy agenda is consistent with Jones’ (1968, pp. 632-634) study of the revolt. An analysis by Schickler (2001, p. 78) also supports this hypothesis.
A second potential motivation flowed from the first. The insurgents challenged the speaker and, in turn, he punished many of them (see Jones, 1968). Norris, for example, saw Cannon as acting in nefarious ways. According to the Nebraska representative, it could happen that an “obstreperous member was given an important committee assignment, which he had long been seeking, and thereby his murmurings were not only quieted, but he became a meek and lowly follower of the Iron Duke.” Or, “if . . . the rebellious member could not be brought back into the reservation by the temptation of political pie, or the fear of political punishment, then he became a political outcast, a good committee assignment for him was impossible.” From Norris’s perspective, “Every individual member who had defied the Speaker’s powers had met with disaster and defeat” (Norris, 1910). Murdock agreed, “The Speaker exercises his power personally in selecting favorites for important committees and punishing others by assigning them to poor committees” (Murdock, 1910, p. 513). Thus, the insurgents had to defeat Cannon to save themselves. There is some evidence that voting against Cannon on key procedural or policy votes in the 61st Congress did reduce the likelihood of a member receiving a place on a requested committee (Lawrence, Maltzman, & Wahlbeck, 2001). Other studies, however, find at best only weak indications that the speaker used committee assignments as a way to punish the insurgents (Krehbiel & Wiseman, 2001, p. 377; Schickler, 2001, p. 78).
Finally, there may have been electoral calculations at play. Norris noted that “Some [insurgents] had openly pledged themselves to vote against [Cannon] in the campaigns which they had recently made” (Norris, 1910). Given the fact that the insurgents largely clustered in a handful of states it seems reasonable to hypothesize that by opposing the speaker and the policies he espoused, they were simply responding to the desires of a majority of their constituents. Indeed, in 1908 a Nebraska paper actually took Norris to task for being too soft on Cannon, “Our present Congressman has voted every session for Joe Cannon for Speaker and for the obnoxious rules of ‘Cannonism’ . . . Voter, you should make Norris pledge himself to vote against Cannon” (McCook Tribune, 1908a). Eventually, Norris and three of his Republican colleagues in the Nebraska delegation made such a promise and their counterparts in Iowa were subjected to similar pressure (Cedar Rapids Evening Gazette, 1908a; New York Times, 1908a, 1908a; Washington Times, 1908b). Democrats even speculated that Republican leaders had given some of their members permission to campaign as being anti-Cannon to help them politically (Commoner, 1908).
By 1910, more anti-Cannon sentiment surfaced, and many newspapers and magazines around the country editorialized against the speaker and his conduct of the House (Atkinson, 1911, pp. 71-93). Thus, in Kansas, “The Eighth district likes Murdock. It likes him because he is against Cannon, because he is against the House rules, because he is a tariff revisionist . . .” (Washington Times, 1909a). In Washington state, a local contact advised Rep. Poindexter, “There is no strong sentiment in this district in [the Speaker’s] favor . . . should you feel constrained to declare yourself on this subject it would . . . be better to be against Cannon” (Allen, 1981, p. 29). These districts were not unusual. In an early 1910 overview of the political problems confronting the GOP, Indiana was said to be “a State where the insurgent or anti-Cannon sentiment is very strong . . . Minnesota is an anti-Cannon State . . . Kansas is practically all insurgent . . . Wisconsin is rampantly against . . . the Cannon machine.” Other states were said to be split between pro- and anti-Cannon forces. In Kentucky with three GOP members, for instance, it was said, “Nobody down there cares whether Cannon is a Czar or a Moujik” (New York Times, 1910a). Thus, some Republican representatives may have found opposing the speaker to be politically easy, even advantageous, while other found it difficult.
For Republicans from districts that had strong anti-Cannon sentiment, if they could make it through their primary, then they might have viewed the revolt as electorally beneficial. Many Republicans faced tough Democratic challenges in 1910. Before the 1910 election, Champ Clark remarked that Cannon was the Democrats’ most valuable asset in the upcoming campaign (Bolles, 1974). Consequently, being against the Speaker could prove advantageous. Baker provides evidence of such political pragmatism on the part of some of the insurgents. He notes that while many of the more progressive rebels were motivated largely by principle, several of the less progressive ones were motivated by electoral needs. They viewed moving against Cannon to be electorally expedient, especially given that many of them faced tough Democratic challenges. Arguably, ideological opposition to Cannonism did not drive these defectors. Indeed, many of these insurgents had supported Cannon on measures that angered their constituents, such as the Payne-Aldrich tariff bill, and they used their participation in the revolt as political cover (Baker, 1973). Again, there is evidence to back up these observations. Notably, Brady and Epstein (1997) elaborate on the constituent differences between the insurgents and other Republicans, finding that the insurgents had disparate policy preferences from other Republicans and received lower-quality committee assignments, which in turn further fueled their desire to rebel. Surprisingly, however, although the possibility that electoral calculations might have motivated those who revolted against Cannon has been suggested before, it has not been subjected to rigorous analysis.
Given these divergent reasons for the insurgency, we do not have a clear picture of the character of the revolt against Cannon or a firm understanding of which of these reasons (or set of reasons) supplied the principal motivation for the insurgents’ actions. Did the insurgents rebel because of their principled stance against Cannon and his policies and behaviors, or for more pragmatic reasons, such as electoral need, or in retribution for Cannon’s despotic control of the chamber, or some combination of all these reasons? We systematically explore which variables carried the most weight in explaining which Republicans revolted against Cannon. This analysis provides insight into why the insurgents revoked their allegiance to the Speaker and their party.
Data and Methods
To explore the nature of the revolt against Cannon, our data include characteristics of the Republican members of the 61st Congress. We consider two dependent variables. First, we examine the members who were included in the Washington Herald’s list of insurgents in January 1910; we refer to these members as early rebels. The Herald produced the list as part of a larger story titled, “First Authoritative Statement as to ‘Insurgent’ Aims in Congress” (Washington Herald, 1910). This dependent variable provides us with information on those members who were thought to be dissidents prior to the vote on the Norris resolution and affords us the opportunity to gain insight into the differences between the members considered rebels in January 1910 and the members who actually rebelled three months later. Second, we examine the Republicans who are customarily characterized as insurgents in the revolt against Cannon. That is, the GOP representatives who voted for Norris’s resolution to remove the speaker from an elected and enlarged Rules Committee. We refer to these members as insurgents.
We divide the early rebels and the insurgents into progressive and nonprogressive early rebels or insurgents; thus, our dependent variables have three categories, that is, progressive early rebel or insurgent, nonprogressive early rebel or insurgent, and nonearly rebel or noninsurgent. This ideological classification is post hoc, based on the members’ participation in the National Progressive Republican League in January 1911, except for Eben Martin (Republican, SD) who is coded as a progressive because he ran as a Progressive Party candidate in 1912. 1 We divide the dependent variables into these categories because we suspect that the driving factors for early rebellion and/or insurgency might be different for these disparate groups. The insurgents who voted against Cannon in March 1910 were never considered a monolithic faction (Baker, 1973; Bolles, 1974, p. 174; Nye, 1959, pp. 247-248). The progressive insurgents voted together on measures before and after the revolt. The nonprogressive insurgents, however, did not have a great deal of cohesion with the progressive insurgents, either before or after the Norris resolution. As Cannon himself observed, insurgents as a group were held together only by anti-Cannon hostility, a point on which Norris agreed (Cannon, 1927; Norris, 1961, p. 122). Given the distinctions among them, there might be motivational differences between the progressives and nonprogressives in both the early rebel and insurgent groups. There were 28 early rebels (12.8% of the Republicans in the 61st Congress), 14 progressives and 14 nonprogressives and 42 insurgents (19.2% of Republicans in the 61st Congress), and 16 progressives and 26 nonprogressives (Baker, 1973). Given that each of our dependent variables has three unordered categories, we use multinomial logistic regression to estimate our models. The base outcome for the multinomial logit models is nonearly rebels for the early rebel models and noninsurgents for the insurgent models.
In order to test the diverse explanations for the insurgency, we include independent variables that capture the hypothesized motivations of the early rebels and insurgents. First, to explore the relationship between the policy and ideological differences with Cannon and participation in the revolt, we include a measure of the member’s ideological distance from Cannon. This is captured using the difference between a member’s DW NOMINATE score and Cannon’s score, with more liberal Republicans taking positive values and more conservative members taking negative values. 2 If divergent policy preferences and ideological principles motivated insurgents, then the ideological distance from Cannon should be positively related to whether a member rebelled against Cannon.
Second, we examine whether Cannon’s heavy-handed rule and his willingness to use his position on the Rules Committee and committee assignments to control the chamber contributed to the rebellion. We consider whether frustrated legislative agendas or poor committee assignments drove the insurgents to rebel. To capture Cannon’s role in obstructing a member’s legislative agenda, we include the percentage of public bills sponsored by a member that were reported from the committee of jurisdiction but never made it out of the Rules Committee in the 60th Congress. We use the 60th Congress because this time frame preceded any major acts of insurgency. As mentioned above, at the beginning of the 61st Congress before committee assignments were made, a group of early insurgents attempted to limit Cannon’s power through rule changes and voted against Cannon on the Payne–Aldrich Tariff. Thus, any subsequent punishment (e.g., blocked bills or poor committee assignments) might not actually be a driving factor in the insurgency but simply a consequence of early defection. That is, retaliation for these sanctions might not be a motive of the insurgency but instead the insurgency might be a cause of the sanctions. Because of this blurred line of causation, we use the percentage of bills that were blocked by the Rules Committee in the 60th Congress to capture policy obstruction by Cannon prior to any major acts of rebellion. If the motivation behind the insurgency was Cannon’s undemocratic use of agenda control, then this variable should be positively related to whether a member revolted against Cannon.
Because we use the percentage of bills blocked in the 60th Congress, we exclude first-year members of the 61st Congress when we include this variable in a model. In order to present models that include percentage of bills blocked in the 60th Congress and also present models that include first-year members of the 61st Congress, we present two models for each of our dependent variables—one that includes the percentage of bills blocked in the 60th Congress, and, thus, excludes first-year members in the 61st Congress and one that excludes the percentage of bills blocked in the 60th Congress, and, thus, includes first-year members in the 61st Congress.
For committee assignments, we employ two separate measures. The first is the difference in a member’s highest valued committee from the average highest valued committee of the member’s peers (based on seniority) in the 61st Congress. 3 We use the committee values estimated by Krehbiel and Wiseman (2001). To create this measure, we subtract the average highest committee value of a member’s peer group from the member’s highest committee value (i.e., member i’s highest committee value—mean highest committee value of peer group). 4 Thus, a lower value indicates a lower-ranked committee assignment relative to the average committee assignment of a member’s cohort. 16 If one of the motivations behind the revolt was retribution for Cannon’s discriminatory assignment of committee seats, then this variable (highest committee value) should be negatively related to whether a member opposed Cannon. The second committee measure is whether a member was denied his committee request or was removed from his previous committee assignment without compensation in the 61st Congress. 5 To determine whether members were denied their committee requests, we use Cannon’s personal notebooks in which he lists the committee requests for the 61st Congress. 6 Given that not all Republicans were recorded as making committee assignment requests in the 61st Congress, we also code whether a member was removed from his previous committee without compensation. If members were denied their committee requests or removed from a previous committee assignment without recompense, then they are coded as 1 and 0 otherwise. If more than one committee request was made, we code a member as 0 if he received one of his requests. If the motivation behind the insurgency was retribution for Cannon’s unjust allocation of committee assignments, then this variable should be positively related to whether a member revolted against Cannon.
Unfortunately, for the committee assignment measures, we are unable to avoid the type of causal uncertainty discussed above by using an earlier Congress. For committee requests, the data from the 60th Congress are unavailable, and for committee portfolio values, it is unclear how to code members who were not a part of the 60th Congress. This, however, does not mean that these variables cannot provide an important piece of the story. Despite their potential link to early insurgency, poor committee assignments may have deepened the early insurgents’ desire to oust Cannon and, therefore, represent a significant motivation for the rebellion in March 1910. Also, for those who were not part of the early insurgency and received poor committee assignments for other reasons, these committee assignments might have inspired their defection. Just as important, the link between committee assignments and the insurgency tells us whether the rebellion was mainly comprised of members with poor committee assignments. Was Cannon able to quell the seditious motivations of potential insurgents with good committee assignments? Thus, while the committee assignment measures cannot necessarily tell us whether poor committee assignment motivated the rebellion, they are, at the very least, important controls and inform us of the relationship between committee assignments and rebellion.
Finally, to investigate the possibility of electoral calculations motivating the rebellion, we include measures of the political environment of the members’ home states and their electoral vulnerability. The measure of electoral vulnerability is a dummy variable that equals 1 if a member’s electoral margin in 1908 was less than or equal to 5 percentage points and 0 otherwise. To capture the level of progressive/populist sentiment in a member’s state, we include the percentage of lower-house members of the state legislature who were elected on the Populist, Fusion, or Independent Republican (IR) party tickets between 1890 and 1908, starting at the height of the populist movement and ending at the election before the insurgency. 7 We use these data because although there were important nuances differentiating their political theories and bases of appeal, the insurgents and their Populist predecessors were focused on similar sets of policy issues, including greater regulation of business and workers’ rights, and a similar desire to eliminate undemocratic practices in government (Argersinger, 1980, pp. 303-305; Hechler, 1940, pp. 21-22; Hofstadter, 1955, pp. 133-134; Sanders, 1999, pp. 158-160). Indeed, when describing the 1906 elections in Nebraska, a state from which a number of insurgents hailed, Lowitt (1963, p. 114) observes that “several of the Republican candidates sounded like old-time Populist orators.” To examine the relationship between electoral need and the probability of rebelling against Cannon, we interact the measure of electoral vulnerability and the level of support for progressive ideals in a state. If there is an association between electoral forces and the likelihood of rebelling, then the interaction effect should be positive and significant, indicating that if members from states with strong progressive ties are electorally vulnerable, then they are more likely to join the insurrection against Cannon.
We also control for the seniority of the members, measured as the number of consecutive terms served. Less senior members are more likely to undermine the norms of the chamber and participate in activities that challenge the authority of congressional leaders, such as discharge petitions (Miller & Overby, 2010; Pearson & Schickler, 2009). Thus, more senior members may have been less likely to rebel against the speaker and to work to alter House rules.
Findings
Before showing the results of the multivariate analyses, we present some descriptive statistics that provide insight into the character of the revolt. Table 1 presents the differences between all the early rebels, the insurgents, and all other Republicans for ideological preferences, seniority, electoral security, the percentage of bills blocked by the Rules Committee, and committee assignment values. Looking first at the ideological distinctions between the groups of early rebels and nonearly rebels and insurgents and noninsurgents, the progressives (both early rebels and insurgents) were more liberal relative to Cannon than the nonprogressives (both groups) and the nonearly rebels/insurgents. This comports with the notion that the progressive members of the rebellion (early rebels and insurgents) had strong policy differences with Cannon and that their opposition to him had some foundation in principle. The nonprogressive early rebels and insurgents were more liberal than the nonearly rebels/insurgents, though not quite to the extent of the progressives. The nonprogressives (both groups) were less senior than the progressives and nonearly rebels/insurgents, and these members (both groups) had a greater percentage of bills blocked by the Rules Committee than their progressive and nonearly rebel/insurgent colleagues.
Differences Among Republican Members on Ideology, Seniority, Electoral Margins, and Committee Portfolio Values.
Difference between progressives and all other members significant at p ≤ .10, two-tailed test.
Difference between nonprogressives and all other members significant at p ≤ .10, two-tailed test.
Difference between progressives and nonprogressives significant at p ≤ .10, two-tailed test.
Excludes first-year members of the 61st Congress
Does not include Cannon (IL), Cassidy (OH), Lovering (MA), or Perkins (NY)
While similar patterns emerge between the insurgents and the early rebels for ideological distance, seniority, and blocked policy agendas, a discrepancy between these two groups materializes in electoral margins and the distance from the average highest committee value. For electoral margin, while nonprogressive early rebels were less electorally vulnerable than their progressive and nonearly rebel counterparts, the reverse is true for the insurgent group. In the 1908 elections, 19.2% of the nonprogressive insurgents had electoral margins of 5 percentage points or less, while only 12.5% of the progressive insurgents were in this category and 16.7% of all other Republicans. These differences are, however, statistically insignificant. Conversely, many of the progressive insurgents were relatively safe in 1908. Indeed, 63% of the progressive insurgents enjoyed electoral margins greater than 15 percentage points in 1908, while only 35% of nonprogressive insurgents and 53% of all other Republicans had electoral margins of this size. The differences between progressives and nonprogressives and all Republicans and nonprogressives are statistically significant. 8
For committee assignments, generally, early rebels and insurgents have worse committee assignments than other Republicans. However, the nonprogressive early rebels have lower committee values compared to their cohort than the progressive early rebels, while the reverse is true for the insurgents. This suggests that committee assignments were not driving the decision to revolt for the nonprogressive insurgents, at least not for those who joined the insurgency after January 1910. Many of the nonprogressive insurgents received decent committee assignments. James Davidson (Republican, WI), for example, was reassigned to Rivers and Harbors and remained chair of Railways and Canals, despite voting against Cannon on the previous question on readopting the rules from the 60th Congress, the vote on adopting the rules from the 60th Congress for the 61st Congress, the previous question on Clark’s proposal to enlarge the Rules Committee, and the Fitzgerald rules compromise providing for Calendar Wednesday. Even more telling, not one of the six nonprogressive insurgents who received committee assignments valued at greater than 1, including one assigned to Appropriations (Edward Taylor, Republican, OH) and one to Judiciary (L. Paul Howland, Republican, OH), were part of the early Washington Herald rebel list.
A similar pattern emerges in Table 2, which presents the percentage of insurgents, early rebels, and nonearly rebels/insurgents who were denied committee seats as well as the percentage of members who voted against Speaker Cannon on a few key votes. Among the insurgents, a lower percentage of nonprogressive insurgents were denied committee assignments than progressive insurgents, although, among the early rebels, 50% of the members of each group were denied their committee requests or lost previously held seats. The denial of committee requests is linked to members’ votes on a few key issues (Lawrence et al., 2001), which is clear when looking at the percentages of insurgents and noninsurgents who were denied committee placements. Also, when comparing the early rebels to the insurgents, it is evident that the former group was more consistent in their opposition toward Cannon. All of the early rebels voted against the speaker on at least one of the five key votes that Cannon considered very important, which were the speakership vote, the previous question on readopting the rules from the 60th Congress, the vote on adopting the rules from the 60th Congress for the 61st Congress, the previous question on Clark’s proposal to enlarge the Rules Committee, and the Fitzgerald rules compromise providing for Calendar Wednesday. Also, a greater proportion voted against the call for the previous question for the rule and/or the rule for the Payne–Aldrich tariff bill. Many of the early rebels, progressive and non progressive, had campaigned in 1908 on the evils of Cannonism; they came to the 61st Congress on a mission to oppose the speaker, and they did so right from the start of the first session.
Differences Between Republican Members on Committee Requests, Killed Bills, and Key House Votes.
Difference between progressives and all other members significant at p ≤ .10, two-tailed test.
Difference between nonprogressives and all other members significant at p ≤ .10, two-tailed test.
Difference between progressives and nonprogressives significant at p ≤ .10, two-tailed test.
In sum, these numbers indicate that progressive early rebels and insurgents were more ideologically distinct from Cannon, suggesting a more principled stance against the speaker. Additionally, the nonprogressives insurgents were more electorally vulnerable than their progressive counterparts, which may be a clue to their motivations for rebelling against Cannon. Also, given that many of the nonprogressive insurgents received high-quality committee assignments, retribution for unjust committee assignment does not seem to be their motivation for rebelling, at least not retribution for their own committee assignments.
These patterns are further illuminated in our multivariate analyses. Table 3 presents the models exploring the motivations of the early rebels. As mentioned above, because of the coding of the percentage of bills killed by the Rules Committee, we present two models—one that includes the percentage of bills blocked in the 60th Congress and excludes first-year members in the 61st Congress (Model 1) and one that excludes the percentage of bills blocked in the 60th Congress and includes first-year members in the 61st Congress (Model 2). In both models, the strongest indicator of whether a member is included on the Washington Herald’s list of early rebels is a member’s ideological distance from Cannon. As members move away from Cannon on the ideological spectrum in a more liberal direction, the more likely it is that they were listed as early rebels. In Model 1, moving from the minimum (−0.423) to maximum value (0.382) of ideological distance from Cannon increases the probability of a member’s inclusion on the Washington Herald’s list as a progressive early rebel by 0.99 (p < .01) and increases the probability of a member’s inclusion as a nonprogressive early rebel by 0.01. 9 When excluding the percentage of bills blocked in the 60th Congress (Model 2), shifting from the minimum (−0.423) to maximum (0.471) value of ideological distance from Cannon increases the probability of a member’s inclusion on the Washington Herald’s list as a progressive early rebel by 0.97 (p < .01) and increases the probability of a member’s inclusion as a nonprogressive counterpart by 0.03.
Motivations of Early Rebels in the Revolt Against Cannon.
Note: The base outcome for the multinomial logit models is members coded as 0 for early rebel.
p < 0.01, ** p < 0.05, * p < 0.10, two-tailed test
Additionally, a member’s highest committee value significantly affects whether a member is a nonprogressive early rebel in Models 1 and 2. The lower a member’s highest committee value relative to that of his peers increases the likelihood that he will be a nonprogressive early rebel. While the probability change based on the estimates from Model 1 is insignificant, when considering Model 2, the probability that a member is designated as a nonprogressive early rebel increases by 0.25 (p < .03) when moving the highest committee variable from its maximum (1.75) to its minimum (−1.18). 10 This provides evidence that retribution for committee assignments may have been a factor in the decision to challenge Cannon for nonprogressive early rebels. But we cannot rule out the possibility that nonprogressive early rebels were punished for their early opposition to Cannon, leading relatively low committee values to be correlated with their early insurgency.
More important for our purposes is the null finding for both committee assignment measures for progressive early rebels in Models 1 and 2. This finding indicates the neither high nor low committee values are linked to early rebellion, suggesting that even the progressive early rebels who were not punished in their committee assignments still participated in the early opposition. These members were not bought off when Cannon did not reduce their committee status, which suggests that committee retribution, at least personal reprisal, was not a key motivation of the progressive early rebels.
Also, in Models 1 and 2, the interaction between electoral vulnerability and the percentage of state lower chamber members elected as Populists, Fusionists, or IR is insignificant for the progressive early rebels, indicating that electoral need is not associated with whether these members are listed as early rebels regardless of the support for progressive ideals in their state. The interaction is significant for nonprogressive early rebels in Models 1 and 2, although the probability changes based on the estimates from these models are insignificant. 11 These findings indicate that while nonprogressive early rebels might have included electoral need in their decision to oppose Cannon, the substantive effect is not robust. 12
Additionally, in Models 1 and 2, one of the interaction components, the percentage of state lower-chamber members elected as Populists, Fusionists, or IR, positively affects the likelihood that a member is listed as a progressive early rebel. This is the effect of the percentage of state lower-chamber members elected as Populists, and so on when the other interaction component, close election, equals 0, indicating that for members who were less electorally vulnerable, support for progressive ideals in their home state increases the likelihood that these members were listed as progressive early rebels. But the minimum to maximum probability changes are insignificant. The other interaction component, close election, has a negative effect on the likelihood that a member is listed as a nonprogressive early rebel in Model 1 and a progressive early rebel in Model 2. This indicates that when the percentage of state lower chamber members elected as Populists, and so on equals 0, electoral vulnerability is negatively associated with inclusion on the nonprogressive early rebel list (Model 1) and inclusion on the progressive early rebel list (Model 2). Again, however, the minimum to maximum probability changes are insignificant.
The analysis of the early rebels presents a picture of the initial insurgents, the ones who were consistently in opposition to Cannon, and lays a foundation for the examination of the insurgents who voted in favor of the Norris resolution on March 19, 1910. These insurgents were a larger group and included some members who had not defied Cannon before that day. The results of the models predicting the members who voted in favor of the Norris Resolution (insurgents) are presented in Table 4, with one model that includes the percentage of bills blocked in the 60th Congress and excludes first-year members in the 61st Congress (Model 3) and one model that excludes the percentage of bills blocked in the 60th Congress and includes first-year members in the 61st Congress (Model 4). As in the analysis of the early rebels, more liberal Republicans relative to Cannon were much more likely to revolt against him, although the relative probability change is only significant for progressive insurgents. In Model 3, as ideological distance shifts from its minimum to its maximum value, the probability of a member rebelling as a progressive insurgent increases by .89 (p < .01) and the probability of rebelling as a nonprogressive insurgent increases by .11; in Model 4, the probability of rebelling as a progressive insurgent increases by .92 (p < .01) and the probability of rebelling as a nonprogressive insurgent increases by .08. This reveals that there were policy differences between the insurgents and Cannon, and that perhaps the insurgents’ behavior was, at least in part, driven by principle.
Motivations of Insurgents in the Revolt Against Cannon.
Note: The base outcome for the multinomial logit models is members coded as 0 for insurgent.
p < 0.01, ** p < 0.05, * p < 0.10, two-tailed test
For the nonprogressive insurgents, however, there are additional agents at work. The inability to move an agenda forward is linked to the likelihood of rebelling as a nonprogressive insurgent (Model 3). As the percentage of bills blocked by the Rules Committee shifts from its minimum (0) to maximum (33), the probability of revolting as a nonprogressive insurgent increases by .65 (p < .10). This provides evidence that retaliation for the Cannon’s alleged use of the Rules Committee as his personal gatekeeper was a motivating factor for nonprogressive insurgents.
Electoral need is also a strong predictor of whether a member was a nonprogressive insurgent, as indicated by the positive coefficient for the interaction between electoral vulnerability and the percentage of state lower chamber members elected as Populists, Fusionists, or IR in a member’s state in Models 3 and 4. The effect of electoral vulnerability (when close election equals 1) on the likelihood of rebelling is much greater when a member is a product of a political environment that is supportive of the progressive movement. Based on Model 3, the probability of an electorally vulnerable member rebelling as a nonprogressive insurgent as the level of support for progressive ideals in a member’s state increases is presented in Figure 1. While the probability of a vulnerable member participating in the revolt as a nonprogressive insurgent is .007 and is indistinguishable from 0 when the percentage of state lower-chamber members elected as Populists, Fusionists, or IR is 0, when the percentage of state lower chamber members elected from these minor parties is held at its highest level, the probability increases to .59 (p < .03). The probability of participating in the rebellion as a nonprogressive insurgent for electorally insecure members increases by .58 (p < .04) as the percentage of state lower-chamber members elected as Populists, Fusionists, or IR increases from its minimum to maximum value. When excluding the blocked bills variable and including first-year members (Model 4), the substantive effects, however, are insignificant. 13 Thus, we find that when controlling for blocked policy agendas and excluding first-year members, revolting against Cannon as a nonprogressive insurgent had significantly greater appeal if a member was electorally vulnerable and hailed from a state that supported progressivism.
Probability of an electorally vulnerable member rebelling as a nonprogressive insurgent.
This finding suggests that in addition to policy differences and retaliation for impeded policy agendas, electoral calculations were a motivation in the decision to rebel for nonprogressive insurgents. To this point, many of the nonprogressive insurgents, such as L. Paul Howland (Republican, OH), made the revolt against Cannon a central piece of their 1910 election campaigns and used their participation in the revolt as cover for their support of the Payne–Aldrich tariff bill as well as other measures (Baker, 1973). This speaks to the importance of electoral concerns for the nonprogressive insurgents who ultimately decided to vote in favor of the Norris resolution, whether they had been regular antagonists of Cannon or not.
More important, electoral considerations and blocked policy agendas are not linked to the likelihood of participating in the rebellion as a progressive insurgent. The coefficient for the percentage of bills killed by the Rules Committee is insignificant, showing that this form of retaliation was not a key motivator. Also, the interaction in the progressive insurgent model is insignificant, indicating that the effect of electoral insecurity on the likelihood of revolting is not conditional on the level of support for the progressive movement in a member’s state. In other words, electoral concerns were not a leading contributor to the decision to revolt for progressive insurgents. As noted above, most of the insurgents were relatively safe in their districts. As such, while the nonprogressive insurgents might have joined the rebellion to secure their electoral fates, this does not seem to be a primary motivator for progressive insurgents. Regardless of the percentage of lower-chamber members elected from the Populist, Fusion, or IR parties, electoral vulnerability is never significantly associated with the probability of participating in the revolt as a progressive insurgent. Thus, insurgents were not motivated by the prospect of losing to a progressive Democrat; other considerations, such as authentic progressive principles, appear to have been of greater concern when they decided to rebel against Cannon.
The notion that progressive members were not solely driven by political expediency is also supported by the relative stability in membership of the progressive insurgents between January and March 1910 and their relative electoral security in 1910. On one hand, only two additional members joined the progressive group of insurgents between January and March, indicating the intransient nature of the progressive insurgency’s membership. On the other hand, 12 additional members enlisted in the nonprogressive insurgent camp during this time, suggesting that as campaign season approached and nonprogressive Republicans realized that a revolt was going to take place, they calculated the potential electoral benefit of rising up against Cannon when determining whether to join the cause. Also, after the revolt in the 1910 elections, the progressive insurgents had greater electoral margins than the nonprogressive insurgents and other Republicans. Excluding members who did not run for a House seat in 1910, the mean electoral margin for progressive insurgents was 35% compared to 19.3% for nonprogressive insurgents and 9.6% for all other Republicans. 14 All of this evidence supports the idea that while the progressive insurgents presumably benefited electorally from the revolt against Cannon, their immediate electoral fortunes were not the chief reason for their challenge against Cannon.
Going along with the differences in the electoral needs of the progressive and nonprogressive insurgents, the finding for the percentage of lower-chamber members elected from the Populist, Fusion, or IR parties in a member’s state in Models 3 and 4 highlights this distinction. For progressive insurgents, the coefficient for the percentage of lower-chamber members elected from the Populist, Fusion, or IR parties in a member’s state, which represents the effect of this variable when electoral vulnerability equals 0, is positive and significant (Model 4). This suggests that if members were electorally secure, then they were more likely to join the insurgency as progressive members as the level of support for progressive ideals in their state increased, once again underlining the electoral security of these insurgents. The probability change, however, is insignificant. For nonprogressive insurgents, the coefficient for the percentage of lower-chamber members elected from the Populist, Fusion, or IR parties in a member’s state is negative and significant (Models 3 and 4), suggesting that if members were electorally secure, then they were less likely to join the insurgency as nonprogressive members as the level of support for progressive ideals in their state increased. While the probability change based on the estimates from Model 3 is insignificant, the minimum to maximum probability change based on Model 4 is significant (−0.10, p < .01).
Seniority is also related to whether a member rebelled as a nonprogressive insurgent. In Models 3 and 4, more senior members were less likely to join the nonprogressive insurgent camp. While the probability change based on the estimates from Model 3 is insignificant, the probability change based on Model 4 is significant; the probability of rebelling as a nonprogressive insurgent decreases by .13 (p < .04) as seniority increases from minimum (0) to maximum (15).
Also of interest, committee assignments do not significantly affect the likelihood of rebelling as a progressive or nonprogressive insurgent. This suggests that retribution for discriminatory use of committee assignments was not a driving factor in the decision to rebel. Both members who received good and bad committee assignment took part in the insurrection. While these results do not rule out the possibility that members were challenging Cannon because of the unjust way that he doled out committee assignments generally, and in regard to some insurgents, Cannon clearly thought they were motivated by such grievances (Cannon, 1927, pp. 256-257), they do help to eliminate personal committee assignments as the principal motivation for the insurgency. Moreover, the fact that many insurgents, even some early ones, received high-quality committee assignments suggests that good committee assignments did not deflate their opposition to Cannon. 15
In light of the three potential explanations for the insurgency, there are a few broad conclusions that we can take away from these results. First, for progressive early rebels and insurgents, ideological distance from Cannon is the only key variable that has a meaningful influence on the probability of rebelling. This supports the notion that real policy differences drove them to challenge the speaker. Indeed, Rep. Cooper argued on the House floor, “If my motives in attacking the rules had not been sincere, would I have attacked them? I had what ordinarily is called ‘a good thing’ in this House. I had two rooms and a clerk, and I could look as wise as the average chairman does” (Phillips, 1958, p. 169). As mentioned above, many members who held relatively good committee seats still chose to rebel against Cannon. Even Cannon admitted, “Many of the Insurgents were honest and really believed they were the victims of the Speaker and a self-appointed cabal.” He specifically singled out Rep. Gardiner for being “a rebel on principle” (Cannon, 1927, pp. 246, 261). Thus, we find support that the insurgency for the progressive members was founded on principle and policy differences, and the other considerations were less influential. Retaliation for personal grievances (committee assignments or the obstruction of their bills) and electoral incentives do not seem to have been key factors in their insurgency calculations.
However, for the nonprogressive defectors (early rebels and insurgents), the insurgency is characterized by policy differences as well as electoral need and retaliation for personal grievances. Committee assignments are significantly linked to the likelihood of being designated as a nonprogressive early rebel, although whether this is a cause or a consequence of early rebellion is difficult to discern. Also, electoral calculations appear to have influenced nonprogressive early rebels, although the substantive effect is not robust. For nonprogressive insurgents, electoral incentives and retaliation for Cannon’s obstruction of their policy agendas are both driving factors for the rebellion.
Our analysis suggests only a modest role for personal retribution for committee assignments, in contrast to the accounts of Jones (1968) and others who highlight the role that Cannon’s perceived abuse of the committee assignment process played in the revolt. We find that committee assignments are only substantively important for nonprogressive early rebels and do not have a meaningful effect for progressive early rebels or progressive and nonprogressive insurgents. Our findings reflect the fact that many of the early rebels and insurgents were appointed to quality committees. Many of the insurgents, particularly the nonprogressive ones, had not shown any sign of rebellion prior to the Norris vote, and as such, Cannon had no reason to assign them to low-valued committees. And while Cannon took vengeance on a few early rebels who were challenging him from the start of the session, not all early rebels received the same discriminatory treatment, which might have been, as Norris (1910) suggested, an attempt by Cannon to bring these members back into the fold. This, of course, is not to say that members who were unpunished in their committee assignments were indifferent to Cannon’s treatment of other members. As such, our findings do not necessarily minimize the role of Cannon’s punitive use of committee assignments in contributing to the growing resentment of the manner in which the Speaker wielded power in the chamber; they simply indicate that many of the early rebels and insurgents were not motivated by their personal suffering under him. Overall, the impression of Speaker Cannon generated by the results produced here is largely consistent with Krehbiel and Wiseman’s (2001) characterization of him as a majoritarian. Additionally, given that personal grievances, particularly poor committee assignments, do not seem to have been a driving factor for the rebellion, this may also provide a clue to how Cannon treated the minority party. Perhaps, as suggested by Kriehbiel and Wiseman’s (2005) analysis of Cannon’s treatment of the Democrats, bipartisanship was more prominent under the Speaker’s rule than commonly thought.
Conclusion
For the progressive core of the insurgency against Speaker Cannon, our analysis documents that policy differences largely drove their behavior. Although they were disgruntled about committee assignments and the undemocratic manner in which Cannon controlled the legislative agenda, their real grievances were directed at the way the House operated in a broad sense. The core insurgents believed the system was rigged against their policy preferences and that it prevented majorities from working their will. Voters back home seemed to be on their side.
A smaller number of less progressive members later joined the early insurgents. These representatives appear to have backed the effort to topple the speaker for electoral reasons and in retaliation for the obstruction of their bills. They calculated that by the time the Norris resolution came up for a vote, Speaker Cannon was in a weak political position and becoming something of an electoral albatross. They opted to oppose him to make themselves more appealing to the voters back home.
Ultimately, our findings suggest that the different stories about the alleged motivations of the insurgents were all true, just for disparate groups of insurgents. For the progressive insurgents, their insistence that their underlying motivation was founded on their policy differences with Cannon appears to hold merit. This analysis, however, paints a less flattering picture of the motivations of their nonprogressive counterparts. Electoral calculations and getting even for personal grievances seem to be important factors for determining whether they joined the rebellion. Thus, the character of the revolt against Cannon depends on the insurgent.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.