Speedlane improves the load time for all kinds of Internet connections. There is no easy way of measuring the improvement, since transmission times through the Internet vary widely. Measuring programs mostly display the current throughput and do not actually replicate the user's experience. 
To be fair to the user, we are measuring improvements manually by finding the time it takes to totally load one page. Testing procedures take care of emptying the browser's cache, which would otherwise skew the results. 
Testing is typically done at night, because the heavy load on the Internet during the day makes measurements vary too much. 
We invite all users to do similar tests and will post them on this page. 

Test description: We used Windows95 on a 133 MHz Pentium with a 28.8 modem in a lawyers office. We were accessing the Internet at noon through a CompuServe connection with the built-in Internet Explorer. We measured the load time of complete pages until the loading indicator stopped. Between tests the cache was emptied and Compuserve restarted. Each test was run twice and these are the actual values in seconds measured. 
 

Speedlane test results (2) 

Test description: We used a Windows95 laptop at 133 MHz with a Ricochet wireless modem in Palo Alto. To be more independent of network load we waited until 11:00 pm. The measurements were taken with Internet Explorer 4.0. 

Test procedure: 
1) Restart system 
2) Log in to Ricochet 
3) Start Explorer 
4) Delete Cache Files 
5) Load www.realitycom.com until 'Done' appears at bottom 
6) Repeat steps 4-5 once 
7) Use Speedlane to set new value for MTU 
8) Go to 1 
 

MaxMTU	www.realitycom.com (sec)	average	compared to minimum
NOT SET	59 / 59	59.0	47.5 %
100	144 / 209	176.5	341 %
300	69 / 116	92.5	131 %
400	41 / 39	40.0	0 %
500	45 / 46	45.5	14 %
592	48 / 50	49.0	23 %

Result: The Windows95 default MaxMTU value resulted in 47.5 % longer retrieval times than the optimum value. 

Speedlane test results (3) 

Test description: The following tests were done at 3:00 am when the load on the Internet is minimal. We used a Windows95 laptop at 133 MHz with a 33.6 modem. Internet Explorer 4.0. Modem compression was turned off.  The times were taken with the help of NetMedic from Vitalsigns. Three websites (speedlane.com, netscape.com, realitycom.com) were accessed and the load time quoted by NetMedic was noted. Every test was repeated three times and the browser cache was emptied in between trials. 
 

MaxMTU = 800	Test1	Test2	Test3	average	compare to minimum
speedlane.com	31.2	30.0	32.2	31.1	0.0
netscape.com	23.3	20.5	26.0	23.3	0.0
realitycom.com	54.5	43.3	36.2	44.7	34.9
MaxMTU = 576	Test1	Test2	Test3	average	compare to minimum
speedlane.com	31.8	35.1	34.9	33.9	9.0
netscape.com	29.0	33.7	29.9	30.9	32.7
realitycom.com	32.8	33.5	33.0	33.1	0.0
MaxMTU = NOT SET	Test1	Test2	Test3	average	compare to minimum
speedlane.com	36.3	46.8	59.1	47.4	52.2
netscape.com	36.9	36.1	40.3	37.8	62.3
realitycom.com	38.4	41.9	55.7	45.3	37.0
				average:	50.5

The page load times under Windows95 default conditions were up to 62.3 % longer than with the optimum MaxMTU values.