Short Howe Truss Popsicle Bridge

I first designed and built this popsicle stick bridge to complete my short popsicle bridge series (Warren, Pratt, and Howe). This bridge uses the Howe Truss design, which was one of the more common designs for early American truss bridges. It was especially popular when railroads were being built across the nation. I love that we can make a popsicle stick version of this iconic design.

First Version

This first version was 13.5 inches long and used 50 unmodified popsicle sticks. It ended up holding 117 pounds before failing. Here is a video of the testing procedure:


  • Length: 13.5 inches
  • Height: 4 inches
  • Width: 3.5 inches
  • Sticks used: 50
  • Weight supported: 117 lbs

I changed up the testing with this bridge. Instead of placing one very large weight directly on the bridge, I started with a small weight. This increased the concentration of the load in the middle of the bridge. I figured this would be more realistic to the way bridges would be tested when built by other people. This is why I think this bridge held significantly less than the Short Pratt Truss Bridge.

Further Improvements

I have built countless versions of this bridge since the original. Some of these have attempted to push the boundaries of how much weight the bridge can support without failing. Other versions have attempted to make it easy for others to replicate this bridge with success.

In 2024, I conducted a controlled experiment using three sets of three bridges, for nine bridges in total.

Set #1, I called the “base” or the control group. These three bridges are the basic version using 49 sticks for which the blueprints are for sale in my store.

Control Group Characteristics

  • 49 unmodified standard popsicle sticks
  • Very low level sorting (I did not spend much time at all sorting through the popsicle sticks to remove the bent, twisted, cupped, etc. ones). Out of 50 sticks per bridge, I removed about 3 sticks each time
  • Elmer’s Glue All white glue (this glue was fairly old, but seemed to be fine)
  • Followed my blueprints and building instructions exactly as sold in my store
  • Allowed the glue to dry for 48 hours before testing
  • Used the assembly jigs I’ve created for aligning and joining the sides of the bridge
  • Used the Pitsco Bridge tester machine

Second Group Characteristics

  • Everything was the same as the control group except for:
  • 57 unmodified standard popsicle sticks
  • Followed the optional reinforcement instructions as indicated in my building instructions as sold in my store. This adds additional lateral bracing and more sticks to reinforce the top chord.

Third Group Characteristics

  • Everything was the same as the control group except for:
  • Spent more time sorting the sticks into three groups
    • Group 1: Suitable for the top and bottom chords
      • Not twisted or warped as much
    • Group 2: Suitable for truss members
      • No terrible twists, cups, or warps
    • Group 3: Lateral bracing
      • Obvious twists, cups, defects, etc. This is less important when used in this group
  • Most importantly, this group of popsicle sticks was sanded to be as flat as possible wherever they would be glued together. So the top and bottom chord sticks were sanded along their entire length on both sides. The truss members were sanded on one side on both ends

Experiment Hypothesis: Joints are super important

If you couldn’t tell already from reading the 3rd group characteristics, the point of the experiment was to see how important it is to have flat surfaces to glue together. I knew it was important, but I wanted to know exactly how much effect it would have on the amount of weight the bridges could support. What I’ve noticed over the years is that it is almost always a joint failure in these popsicle stick bridges. I wanted to know if the extra time spent sanding the joints would be worth it.

Most popsicle sticks are NOT perfectly flat from the factory. They often have cups, twists, warps, etc. to some degree. Some are worse than others for sure.

Experiment Results

Group #1: Control Group

1st FailurePeak Load
Bridge #1120120
Bridge #28898
Bridge #39191

Group #2

1st FailurePeak Load
Bridge #4135135
Bridge #588135
Bridge #6132185

Group #3

1st Failure LoadPeak Load
Bridge #7158158
Bridge #8118146
Bridge #9142142

Understanding the Results

Here’s a bit more extra information to make sure we are on the same page.

  • 1st Failure Load = the amount of weight on the bridge when the first joint popped loose. There was no other type of first failure other than a a joint coming apart.
  • Peak Load = the amount of weight on the bridge when it experience catastrophic failure.
  • Range = the difference between the least and greatest amount. This points towards consistency.

Control Group Results: As Expected

Group One: Average Peak Load

The control group performed just as I expected based on my previous tests. I haven’t recorded the tests as rigorously in the past as I did in this experiment, but I do have written records scattered about that confirm all these were in the range with no surprises.

I do want to highlight that for two of the bridges, the peak load did not exceed the first failure load. The one bridge that did continue on only gained 10 pounds extra. This was different than the other two groups.

Group Two (Reinforced): Increase in strength

Group Two: Average Peak Load

This group was expected in increase in strength with the added reinforcements. It did not disappoint in that regard. It increased the average first failure point by 18 pounds over the control group, and 32 pounds on the peak load.

Unlike the control group, each bridge was able to continue being loaded after the first failure point. Each first failure point was a tension member joint, mostly along the top chord near the bridge ends.

Interestingly, this group had the highest range of peak load between the bridges. Two bridges had the exact same of 135 lbs and the third had a 185 lbs peak load. While it is never a bad thing for a bridge to over preform, I would prefer there to be more consistency. The range was high in both the peak load and first failure load.

Group Three: Joints Do Matter Tremendously

Group Three: Average Peak Load

Group three proved very clearly that the attention paid to sand down the joints flat, to allow better contact between the wood and glue makes a significantly difference. In fact, when you look at the average peak load, group three did better than group two despite using fewer sticks.

Actually, group three did better in Min Peak Load, Average Peak Load, and Range out of all groups. If it weren’t for the odd super strong bridge in group two, group three probably would have beaten them out in everything. I especially want to point out the consistency, with only a range of 16 in peak load. Bottom line, sanding the joints made a huge difference for this experiment.

This will become a prominent part of the methodology for my own popsicle stick bridges in the future and be incorporated into my building instructions as well.

Ideas for Further Experiments

I bet that if I sanded the joints for Group Two, the reinforced version, it would show a similar improvement as Group One to Group Three and would become the best performing group by far.

I also bet that this experiment has a direct applicated to all popsicle stick bridges. In my experience of using tens of thousands of popsicle sticks (not an exaggeration), it is hard to find ones that create perfect joints.

Along those lines, I think it is worth experimenting with a “no glue” joint, such as a pin joint to completely remove the weak point and inconsistency of a glued popsicle stick joint.

Where to Purchase Plans and Kits

Please visit my store if you are interested in building this exact bridge.

14 thoughts on “Short Howe Truss Popsicle Bridge”

  1. There is a cut in the video at 0:27
    I am making a bridge with a limited number of Popsicle sticks (60) and it has to be at least 1.9 inches tall and 1.9 inches wide (or 5 centimetres, if you live in canada like me)

  2. Hey. I guess I SUPER underestimate Popsicle sticks. I never thought that something like that could hold up to 117lbs but I guess jokes on me. I’m doing a Popsicle stick bridge project myself ( with a group of 3 others) and ,of course, there are bridge specs: 24″ minimum length, 300 Popsicle stick max limit , use carpenter’s glue only, at least one popsicle stick wide and will be placed across two desks spaced out by 20″.
    We’re planning on making it an arch bridge modeled after the Royal Gorge bridge. We already have a general Idea on how to make it with the exception of the width. With the size , Popsicle stick limit ,and the space between desks, how wide would you think would be a good median between strength and the least Popsicle sticks used?
    Also, does cross-bracing help support the weight, or is it even worth the extra sticks used?

    • Zachary,

      You don’t want to make it any wider than you have to. In fact, I would wonder if you could make it just as wide as the bridge in this article. It is “almost” one popsicle stick wide. Otherwise in order to connect the sides of the bridge you will have to use double the popsicle sticks.

      If by cross bracing you mean lateral bracing, then yes. It is definitely worth it.

      • Ok. Thanks for da tip. I have one more question for you. What website could I go to actually “build” a “clone” of my bridge at school and virtually “test” it so maybe I can use that info to improve my bridge without building a separate one at my house that’s more or less identical? It’ll save me a butt-load if time. Thanks in advance 🙂

    • I am pretty sure I used Weldbond for glue on this bridge. I didn’t use anything other than popsicle sticks and glue for materials in this bridge. I had a couple other things to help me make it, such as the blueprints and some heavy books.


Leave a Comment